ABCDEFG

UNIVERS ITY OF OULU P .O . Box 7500 F I -90014 UNIVERS ITY OF OULU F INLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

S E R I E S E D I T O R S

SCIENTIAE RERUM NATURALIUM

HUMANIORA

TECHNICA

MEDICA

SCIENTIAE RERUM SOCIALIUM

SCRIPTA ACADEMICA

OECONOMICA

EDITOR IN CHIEF

EDITORIAL SECRETARY

Professor Mikko Siponen

Professor Harri Mantila

Professor Juha Kostamovaara

Professor Olli Vuolteenaho

Senior assistant Timo Latomaa

Communications Officer Elna Stjerna

Senior Lecturer Seppo Eriksson

Professor Olli Vuolteenaho

Publication Editor Kirsti Nurkkala

ISBN 951-42-7988-3 (Paperback)ISBN 951-42-7989-1 (PDF)ISSN 0355-3213 (Print)ISSN 1796-2226 (Online)

U N I V E R S I TAT I S O U L U E N S I SACTAC

TECHNICA

OULU 2006

C 235

Jari Juuti

PRE-STRESSED PIEZOELECTRIC ACTUATOR FOR MICRO AND FINE MECHANICAL APPLICATIONS

FACULTY OF TECHNOLOGY,DEPARTMENT OF ELECTRICAL AND INFORMATION ENGINEERING,INFOTECH OULU,UNIVERSITY OF OULU

C 235

AC

TA Jari Juuti

C235etukansi.fm Page 1 Friday, March 3, 2006 3:24 PM

A C T A U N I V E R S I T A T I S O U L U E N S I SC Te c h n i c a 2 3 5

JARI JUUTI

PRE-STRESSED PIEZOELECTRIC ACTUATOR FOR MICRO AND FINE MECHANICAL APPLICATIONS

Academic Dissertation to be presented with the assent ofthe Faculty of Technology, University of Oulu, for publicdiscussion in Raahensali (Auditorium L10), Linnanmaa,on April 7th, 2006, at 12 noon

OULUN YLIOPISTO, OULU 2006

Copyright © 2006Acta Univ. Oul. C 235, 2006

Supervised byProfessor Seppo Leppävuori

Reviewed byDoctor Pasi KallioProfessor Ahmad Safari

ISBN 951-42-7988-3 (Paperback)ISBN 951-42-7989-1 (PDF) http://herkules.oulu.fi/isbn9514279891/ISSN 0355-3213 (Printed )ISSN 1796-2226 (Online) http://herkules.oulu.fi/issn03553213/

Cover designRaimo Ahonen

OULU UNIVERSITY PRESSOULU 2006

Juuti, Jari, Pre-stressed piezoelectric actuator for micro and fine mechanicalapplicationsFaculty of Technology, University of Oulu, P.O.Box 4000, FI-90014 University of Oulu, Finland,Department of Electrical and Information Engineering, Infotech Oulu, University of Oulu, P.O.Box4500, FI-90014 University of Oulu, Finland Acta Univ. Oul. C 235, 2006Oulu, Finland

AbstractIn this thesis pre-stressed piezoelectric actuators for micro and fine mechanical applications havebeen developed. First, RAINBOW (Reduced And INternally Biased Oxide Wafer) and thick filmactuators were manufactured and their electromechanical properties were characterised. In the secondpart, the novel pre-stressed piezoelectric actuator PRESTO (PRE-STressed electrOactive componentby using a post-fired biasing layer) was developed and its electrical and electromechanical propertieswere measured.

Commercial piezoelectric PZT 5A and PZT 5H discs were used in the RAINBOW and PRESTOactuators and PLZT paste for thick film actuators. The pre-stressing of the PRESTO actuators wasbased on the sintering shrinkage and different thermal expansion coefficient of the piezoelectric discand passive material. Dielectric LTCC tape and AgPd paste were utilized as pre-stressing media andpassive layer by using lamination and screen-printing, respectively. Different active and passive layerthicknesses and electrode materials were realized in order to obtain high displacements and good loadbearing capability for actuators.

The PRESTO actuators showed a significantly higher coercive electric field than their bulkcounterparts, but a decreased remanent polarisation. The displacement as a function of load wasmeasured under 0.3–3 N loads and electric fields up to ±0.75 MV/m. The highest displacement of118 μm was obtained with a 250 μm thick PZT 5H actuator (Ø 25 mm) with LTCC tape (thickness~96 μm) as the pre-stressing material. The corresponding actuator with AgPd pre-stressing material(thickness ~33 μm) produced 63 μm displacement. Additionally, PRESTO actuators were tested witha glued steel layer in a mechanical amplifier which obtained displacements up to 1.2 mm.

Effective d31 coefficients of the PRESTO actuators were derived using an analysis based onunimorph model and measured displacement data. The actuators exhibited significantly enhancedeffective d31 coefficients (d31eff = -690 pm/V and d31eff = -994 pm/V for PZT 5A and 5H,respectively) comparable to the RAINBOW actuators. Mass-producible PRESTO actuators with highdisplacement, moderate load bearing capabilities and integration possibilities can be utilised invarious micro and fine mechanical devices e.g. dosing devices, electromechanical locks, regulators,positioners vibrators, speakers, adjusters, pumps, valves, relays, dispensers, micromanipulators, etc.

Keywords: actuator, piezoelectric, pre-stress, PRESTO, RAINBOW

“Simplicity is the key to brilliance” Bruce Lee

Dedicated to my family and friends

Acknowledgements

First of all, I wish to thank my supervisor Professor Seppo Leppävuori, who arranged equipment, facilities and other circumference fertile for profound research. I am also most grateful to Professor Heli Jantunen for her support, inspiration and mental tutoring of this work. I wish to thank also Dr. Antti Uusimäki for revising this work and guiding me in research work in piezoelectric ceramics together with M.Sc. Hannu Moilanen. I am especially grateful to colleagues in the Microelectronics and Material Physics Laboratories for the cosy working atmosphere. Special thanks are dedicated to Mr. Pekka Moilanen, Mr. Antti Paavola, M.Sc. Esa Heinonen and Dr. Krisztian Kordas for their tireless help and support in many ways. I am very grateful to Professor A. E. Hill for revising the English language of the manuscript.

This work was financially supported by the Academy of Finland (TUKEVA program), TEKES (PRESTO program), Jenny and Antti Wihuri Foundation, Ulla Tuominen Foundation, Tauno Tönning Foundation, Infotech Oulu Graduate School and Electronic Materials, Packaging and Reliability Group of Infotech Oulu. All are gratefully acknowledged.

Special thanks are expressed to my friends who have been patient with this one topic in my recent life. I also thank my parents and parents-in-law, little sister and her family and little brother for their support. Finally, my deepest gratitude is dedicated to my dear and patient wife and son, Marja and Juho Juuti, those who have shared storms and sunshine with me in varying stages of life.

Thank you all!

Oulu, February 2006 Jari Juuti

List of abbreviations and symbols

δ Displacement along the longitudinal axis ν Poisson’s ratio νnp Poisson’s ratio of non-piezoelectric material νp Poisson’s ratio of piezoelectric material Δhmax,circ The maximum calculated axial displacement of the circular unimorph

actuator Δhmax,rect The maximum calculated axial displacement of the rectangular

unimorph actuator Δhmeas Measured axial displacement d31 Piezoelectric coefficient at 90° from direction of polarisation d31eff Calculated effective piezoelectric coefficient at 90° from direction of

polarisation d33 Piezoelectric coefficient in direction of polarisation dij Piezoelectric coefficient; electric field applied at i axis and strain in j

axis E Electric field Ec Coercive electric field Enp Young’s modulus of non-piezoelectric material Ep Young’s modulus of piezoelectric material f Frequency F Force Ly Width of an actuator Lx Length of an actuator P Vector of poling direction Pr Remanent polarisation

1S Strain along longitudinal axis tnp Thickness of a non-piezoelectric material tp Thickness of a piezoelectric material V Voltage Cermet Composition of ceramic and metal CTE Coefficient of Thermal Expansion

CERAMBOW CERAMic Biased Oxide Wafer FEM Finite Element Method FIB Focused Ion Beam LaRC-SI™ Amorphous thermoplastic developed at NASA LIPCA LIghtweight Piezo-composite Curved Actuator LTCC Low Temperature Co-fired Ceramic MEMS Micro-Electro-Mechanical System PRESTO PRE-STressed electrOactive component by post-fired biasing layer PZN-PZT Composition of a Lead Zinc Niobium and Lead Zirconate Titanate PZT Composition of a Lead Zirconate Titanate PLZT Composition of a Lead Lanthanum Zirconate Titanate RAINBOW Reduced And INternally Biased Oxide Wafer RB RAINBOW THUNDER THin-layer composite UNimorph ferroelectric DrivER and sensor

List of original papers

Original papers, are referred to throughout the text by their roman numbers.

I Juuti J, Heinonen E, Moilanen V-P & Leppävuori S (2004) Displacement, stiffness and load behaviour of laser-cut RAINBOW actuators. Journal of the European Ceramic Society 24: 1901-1904.

II Juuti J, Lozinski A & Leppävuori S (2004) LTCC compatible PLZT thick-films for piezoelectric devices. Sensors and Actuators A 110: 361-364.

III Juuti J, Jantunen H, Moilanen V-P & Leppävuori S (2006) Manufacturing of pre-stressed piezoelectric actuators by post-fired biasing layer. Accepted for publication to IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control in December, 2005.

IV Juuti J, Jantunen H, Moilanen V-P & Leppävuori S (2005) Poling conditions of pre-stressed piezoelectric actuators and their displacement. Journal of Electroceramics 15: 57-64.

V Juuti J, Kordás K, Lonnakko R, Moilanen V-P & Leppävuori S (2005) Mechanically amplified large displacement piezoelectric actuators. Sensors and Actuators A 120: 225-231.

Paper I describes RAINBOW actuator manufacturing from commercial PZT 5A ceramic. Disc samples were laser-cut to “rectangular” shape. The effect of the cutting on dielectric, mechanical and electromechanical properties was studied. Results demonstrate that pre-stressed actuators maintain their mechanical bias after cutting and provide reasonable load bearing capabilities. This shows potential for application specific modifications of actuator properties and shapes.

The topic of Paper II was to introduce a low temperature piezoelectric PLZT paste compatible with the LTCC process. It introduces powder and paste preparation and sample manufacturing via screen-printing on alumina surfaces through firing at 850 °C. Hysteresis loop and electromechanical properties were measured from pre-stressed cantilever samples with alumina substrates.

Papers III and IV introduce development and characteristics of novel PRESTO actuators. Paper III describes manufacturing of the pre-stressed PZT 5A and PZT 5H

actuators by a post-fired biasing layer with different configurations using AgPd paste and LTCC tape as a pre-stressing material. Results show the effects of the constraining top electrode and the thickness of the pre-stressing layer. Electromechanical properties of the PRESTO actuators under different mechanical loads and electric fields were studied and comparison between RAINBOW and other pre-stressed actuators was made. In Paper IV the effects of the poling conditions for hysteresis loop characteristics were studied for PZT 5A and PZT 5H PRESTO and bulk actuators. Displacement, remanent polarisation and coercive electric field were investigated and comparison with bulk samples and other pre-stressed actuators was made.

Paper V describes the design of the mechanical amplifier for the PRESTO actuators. It also introduces the configuration of a pre-stressed unimorph where the actuator is first pre-stressed and then a steel layer is glued onto it. As the actuators provide relatively high displacement and force they can be utilised via variety kind of simple mechanical amplifiers to obtain very high displacements. The approach of small actuator size with a mechanical amplifier that can be realised with injection moulding is a more cost effective way to reach the required parameters for e.g. fine mechanical applications.

The idea of the PRESTO actuators, manufacturing method, sample manufacturing and measurements of the papers were mostly the contribution of the author. In Papers I and V, simulations/modelling was done in association with the co-authors. Basic ideas were the contribution of the author in Papers I, III, IV, and V. The manuscripts were written by the author with the kind help of the co-authors.

Contents

Abstract Acknowledgements List of abbreviations and symbols List of original papers Contents 1 Introduction ................................................................................................................... 15

1.1 Piezoelectric actuators ............................................................................................15 1.2 Objective and outline of the thesis..........................................................................16

2 Pre-stressed piezoelectric bending actuators ................................................................. 18 2.1 Background.............................................................................................................18 2.2 Characteristics ........................................................................................................20 2.3 Manufacturing methods and performance ..............................................................25

3 Measurement methods and manufacturing of RAINBOW and thick film actuators and their properties ........................................................................................ 33 3.1 Measurement and analysis methods .......................................................................33 3.2 RAINBOW actuators..............................................................................................35 3.3 Thick film actuators................................................................................................41

4 Pre-stressed electroactive actuator PRESTO................................................................. 44 4.1 Idea and principles of the PRESTO........................................................................44 4.2 Manufacturing of PRESTO actuators .....................................................................46 4.3 Properties of the PRESTO actuators.......................................................................47 4.4 Discussion...............................................................................................................55

5 Conclusions ................................................................................................................... 63 References Original publications

1 Introduction

1.1 Piezoelectric actuators

Actuators creating force and displacement are required for a variety of industrial and consumer products to produce regulation, on-off motion or continuous operation. Actuators are an essential part of the machine and are key for better performance and new characteristics of the device e.g. the injection system of the car or an alignment of the optical fibre.

Actuators manufactured from active materials have been under extensive research in the past decade. Piezoelectric actuators produce 100 to 1000 times the mechanical work per unit volume and 10 times the energy per mass compared to conventional actuators (i.e. electromagnetic, hydraulic or pneumatic). Piezoelectric actuators have a wide frequency range, high resolution, fast response times and the ability to generate high forces with low power consumption. (Culshaw 1996)

The direct piezoelectric effect was found by Pierre and Jacques Curie in the 1880s in Rochelle salt and soon after this the converse effect was discovered. The converse piezoelectric effect changes the dimensions of the material under an electric field. After discovery, it took several decades before the phenomena were exploited in commercial applications. The first application was probably ultrasonic detector for submarines which was developed in 1940s during the World War II. (Chopra 2002) Ever since the piezoelectric materials have been developed to improve their performance and extend their feasibility.

The most widely used piezoelectric material today is lead zirconate titanate, PZT, that induces strains of the order of 0.1 % to 0.2 % (Park & Shrout 1997). In order to reach higher displacements, different kinds of actuator configurations are used to amplify dimensional changes of the material. These amplification mechanisms can be divided into externally, frequency and internally leveraged actuators (Niezrecki et al. 2001). Rough estimation of the stress versus strain characteristics of different type of actuators are shown in Fig. 1.

16

Fig. 1. Stress strain relationship of different type of actuators (edited from Haertling 1994 a, Haertling 1997).

Externally leveraged actuators are based on external mechanisms that amplify displacement using e.g. hydraulic amplification, levers or Moonie type structures. (Niezrecki et al. 2001) Frequency-leveraged actuators rely on many small steps typically taken by non-leveraged actuator at high frequency e.g. travelling wave or inchworm motors (Tenzer & Mrad 2004, Uchino 1997, Niezrecki et al. 2001). Internally leveraged actuators include for example unimorph, bimorph, stack and telescope actuators. These actuators produce relatively high displacements utilising bending of the material or the extended length of the piezoelectric material. (Niezrecki et al. 2001)

In order to expand the application range of piezoelectric devices a high/moderate load bearing ability is also needed. The piezoelectric coefficients are a function of the stress bias in the material which can be utilised to enhance the piezoelectric properties of the actuators. Therefore, mechanically pre-stressed internally leveraged actuators such as RAINBOW, THUNDER, etc. have been under research recently offering high displacement and moderate load bearing capability (Fig. 1). These properties, combined with low power consumption, high resolution and fast response times can be utilized in applications of fine and micromechanics, telecommunication, optics, fluidics, medical technology, vibration and noise control, military, space and the car industry.

1.2 Objective and outline of the thesis

The objective of the thesis is to develop an pre-stressing method and internally leveraged pre-stressed piezoelectric bending actuators feasible for integration, mass-production and fine and micro mechanical applications. Such actuators should also be able to produce forces of several newtons and displacements of dozens of micrometres. A further objective was to gather information and performance of the present pre-stressed actuators in order to compare their characteristics.

17

In Chapter 2, characteristics and manufacturing methods for existing pre-stressing actuators are described to give guidelines, requirements and specification for further development and analysis. In Chapter 3, pre-stressed RAINBOW and thick film actuators are manufactured and their characteristics are investigated. In Chapter 4, a new pre-stressed actuator known as PRESTO (PRE-STressed electrOactive component by using a post-fired biasing layer) and its associated manufacturing method is introduced and developed. The objective was to find materials and optimize manufacturing process for these actuators, test their performance and to amplify their displacement with an external fine mechanical amplifier. Theoretical analysis, comparison and further development possibilities are also discussed. Development stages of novel pre-stress piezoelectric actuator are shown in Fig. 2.

Fig. 2. Development stages of novel pre-stressed piezoelectric bending actuator.

In this thesis, a novel mass-producible manufacturing method for pre-stressed piezoelectric bending actuators is introduced. These PRESTO actuators have high displacement, moderate load bearing capabilities and the possibility to be integrated with e.g. LTCC (Low Temperature Co-fired Ceramic) circuit board.

2 Pre-stressed piezoelectric bending actuators

Background information, operating principles, manufacturing methods and characteristics of existing pre-stressed benders are described to give guidelines, requirements and specifications for actuator development. The characteristic properties of the actuators are gathered and theoretically evaluated for further analysis and comparison.

2.1 Background

In the early 1930s, bimorph and unimorph type benders without pre-stress were developed utilising bending of the actuator in displacement amplification. The bimorphs consisted of two active layers working against each other while unimorph type actuators consisted of one active and one passive layer (US Patent 1931, Germano 1971, Niezrecki et al. 2001) In the late 1980s a branch of unimorph type actuators called monomorphs was developed. (Nakamura et al. 1987, Uchino et al. 1987) These utilised a monolithic structure without a separate bonding layer between the active and passive materials. (Uchino 1997) Mechanically pre-stressed internally leveraged bending actuators are categorised as unimorph or monomorph actuators depending on their structure. In such actuators the basic structure is the same, but it is under mechanical pre-stress which alters the actuator behaviour and shape.

The first pre-stressed bending actuator was named RAINBOW (Reduced And INternally Biased Oxide Wafer) and it was introduced by Haertling (Haertling 1994 a) in 1994 (Fig. 3).

19

Fig. 3. Schematic drawing of RAINBOW actuator.

This monomorph actuator structure is able to give load bearing capabilities over 30 N and displacements over 3 mm with a single wafer having 100 mm diameter and thickness of 375 µm. According to Haertling, these actuators produce higher displacements and have greater load bearing capabilities than normal bender actuators due initially to the pre-stressed state of the ceramic. (Haertling 1997) Since then several other pre-stressed actuator structures has been introduced.

The CERAMBOW (CERAMic Biased Oxide Wafer) actuator was introduced by Barron et al. in 1996 and in the same year THUNDER (THin-layer composite UNimorph ferroelectric DrivER and sensor) was introduced by NASA (Barron et al. 1996, Nolan-Proxmire & Henry 1996). The CRESCENT actuator was introduced by Chandran et al. in 1997 (Chandran et al. 1997) and LIPCA (LIghtweight Piezo-composite Curved Actuator) as an improved version of the THUNDER actuator was developed by Yoon et al. in 2000 (Yoon et al. 2003). In these unimorph type actuators active and passive layers were bonded together by using solder or thermally cured adhesive at elevated temperature. Pre-stress is formed during cooling due to the different thermal shrinkage of the piezoceramic and passive material.

An alternative processing route for RAINBOW type piezoelectric actuators was introduced in 1998 by Pearce and Button and in 2001 a d31 gradient flextensional actuator, PrinDrex, was introduced by Li et al. Both actuators utilised a co-firing process in formation of the pre-stress. In such actuators, mismatch in sintering shrinkage, instead of the CTE (Coefficient of Thermal Expansion) difference, is the key to obtaining the pre-stressed state. The actuator developed by Li et al. demonstrated also a functionally graded structure where passive material was replaced by piezoelectric ceramic with a lower d31 piezoelectric coefficient. (Pearce & Button 1998, Li et al. 2001)

The most recent development in the field is a multilayer pre-stressed monomorph actuator via a co-extrusion method, introduced by Yoon et al. in 2005. The method utilises a co-fired composite of PZN-PZT (lead zinc niobium and lead zirconate titanate) and dispersed silver particles introducing the initial pre-stress state due to the CTE mismatch. (Yoon et al. 2005, Yoon et al. 2006) Performance and manufacturing methods of the pre-stressed unimorph-type actuators are introduced in Chapter 2.3 in detail.

20

It should also be noted that other unimorph-type actuators, such as thick and thin film piezoelectric actuators deposited, for example, on silicon, have been produced for a long time (Chen et al. 1995, Lappalainen et al. 1995). Internal asymmetric pre-stress can be also formed in these actuators due to e.g. a different coefficient of thermal expansion or sintering shrinkage. The presence of pre-stress can be recognized in the properties of the structures (Chung et al. 1994, Paradise et al. 1994, Lappalainen et al. 1995, Lappalainen 1999, Leppävuori et al. 1999). Therefore control of the stresses in MEMS (Micro-Electro-Mechanical System) as well as in mesoscale actuators is important in order to improve their characteristics and avoid unbeneficial behaviour or breakage (Shepard et al. 1996, Koch et al. 1998, Iborra et al. 2004).

2.2 Characteristics

In piezoelectric actuators, the displacement generation capability is described with the dij coefficients. These coefficients define the magnitude of the strain or displacement generated under an electric field. The piezoelectric coefficient perpendicular to the poling axis is (Culshaw 1996)

S L1 xd31 VEtp

δ= = (1)

where ( 1S ) is strain and (δ) displacement along the longitudinal axis (x-axis or axis number 1), (E) electric field in the (tp) thickness (z-axis or axis 3) direction, (V) the applied voltage and (Lx) the length of the actuator. Directions of the piezoelectric coefficients parallel (d33) and perpendicular (d31) to the poling axis are shown in Fig. 3. The effects of these coefficients are opposite due to the fact that, when the actuator is excited parallel to the poling field, the thickness of the actuator increases while the length of the actuator decreases. Reverse excitation produces the opposite behaviour. Typically, piezoelectric d33 and d31 coefficients are specified by the material manufacturer under stress-free conditions at low electric fields, constant frequency and/or as a function of temperature. All these factors affect the piezoelectric coefficients.

The behaviour of d33 and d31 under stress depends on material composition (Zhao et al. 1998, Zhang & Zhao 1999, Yang et al. 2000). In Fig. 4 normalised d33 and -d31 piezoelectric coefficients are shown as a function of increasing and decreasing stress. As can be seen, d33 and -d31 piezoelectric coefficients of softer materials (PZT 5A and 5H) tend to decrease when more stress is applied to the ceramic. Therefore, the displacement of an actuator would decrease. On the other hand, harder piezoelectric materials (PZT 4 and 8) show a tendency to increase d33 and -d31 values, which is beneficial in higher load actuator applications. (Zhang & Zhao 1999)

21

Fig. 4. The dependence of weak signal d33 and d31 coefficients of PZT 5H, PZT 5A, PZT 4 and PZT 8 under compressive uniaxial stress parallel to poling (Zhang & Zhao 1999).

In Fig. 5 values of the -d31 piezoelectric coefficient are presented showing the importance of the proper stress level. It can be seen that when stress levels are increased above ~63 MPa piezoelectric materials PZT 5A and 5H have lower |d31| than PZT 4 and 8 (Fig. 5 a). Consequently, values of the |d31| remain decreased after removal of the mechanical pre-stress from PZT 5A and 5H materials (Fig. 5 b). This is due to partial de-poling of the softer piezoelectric materials by mechanical stress while harder materials, PZT 4 and 8, are not as easily de-poled (Zhang & Zhao 1999). Pre-stressing/pre-loading has been utilized, for example, in stacked piezoelectric actuators and high power underwater transducers with resulting improvements in piezoelectric coefficients, electromechanical coupling factors and reliability. In such cases tensile stresses under high electric field driving conditions are kept to a safe level by adjusting the initial compressive stress in the piezoceramic. This is a consequence of the inherently higher mechanical strength of the ceramic under compressive stress than under tensile stress. (Zhang & Zhao 1999, PI Ceramic 2005)

22

Fig. 5. The dependence of weak signal d31 coefficients of PZT 5H, PZT 5A, PZT 4 and PZT 8 under a) increasing and b) decreasing compressive uniaxial stress (Zhang & Zhao 1999). Data extracted from Fig. 4.

In pre-stressed bender actuators, the situation is different because the actuators utilize tensile stress in the lateral direction. Research has been made concerning the effect of the transverse tensile stress on the properties of the bulk and pre-stressed bending actuators. The results show that a suitable stress state will change the initial domain configuration and electromechanical characteristics, especially at high electric fields, leading to an increased d31 coefficient (Li et al. 1997, Schwartz et al. 2001, Schwartz & Moon 2001, Li et al. 2002, Lee et al. 2005).

Mechanical stress also has an effect on the electrical properties of the active ceramic which have been studied in many publications for e.g. bulk and pre-stressed bending actuators. The hysteresis loop (Fig. 6) of the polarisation plays an important role in the investigation of actuator behaviour under a high electric field. (Wang et al. 1989, Wang et al. 1991, Lynch 1996, Dausch 1997, Zhang & Zhao 1999, Ounaies et al. 2001, Zhou et al. 2005) The hysteresis curve provides information about, for example, coercive electric field and remanent polarisation which are crucial in the determination of suitable poling conditions. This is particularly important from the operational point of view, since the stress state in actuators alters the remanent polarisation and coercive electric field that can enhance actuator performance. The coercive electric field (Ec) is the point where polarisation of an actuator is zero (Fig. 6). It determines how high an electric field is required to negate the charge of an actuator due to polarisation of domains in the material. It also provides information about the minimal poling field which should be applied in order to change polarisation and what is the maximal usable electric field during excitation of the actuators. If excitation exceeds the coercive electric field, de-poling or re-poling occurs which has an effect on actuator properties and performance. This is especially important for bimorph actuators where elements are driven with an electric field opposite to poling. Typically, actuators are driven with ±1/3 AC field of the coercive electric field to avoid disturbance of the poling. Similarly, the electric field limit for re-polarisation is about half that of the initial coercive electric field. (Lynch 1996, Hughes & Bednarchik 2000) Using electrical biasing, an actuator can be driven from -1/3 of the coercive field up to 80 % of the maximum allowable tensile strain. By this means much higher AC electric fields can be applied. (Hughes & Bednarchik 2000)

23

Fig. 6. Hysteresis loop of the piezoelectric material.

The remanent polarisation is determined from the intersection of the hysteresis loop and the polarisation axis as the electric field is returned from its maximum and minimum values to zero, as shown in Fig. 6. In other words it describes the polarisation level or density of the remaining charge as a consequence of the orientation of the domains. The magnitude of the piezoelectric properties depends on the polarisation. The level of the remanent polarisation depend on material composition and is non-linearly related to the piezoelectric coefficients e.g. d33 and d31 in polycrystalline piezoelectric ceramics. (Wang et al. 1991, Moulson & Herbert 2003) Therefore, it can be used to optimise the poling conditions and maximise output of the actuators.

In the case of the internally leveraged actuators (Fig. 3) the displacement is mainly produced by bending via the d31 coefficient. Benders are typically thin but long so that the displacement produced by d31 is much larger than that produced by d33 in spite of the fact that d33 is significantly larger than d31 (d33=593 pm/V d31=-274 pm/V for PZT 5H under stress free conditions) (Berlincourt et al. 2005, Morgan Electro Ceramics 2005). Displacement of the actuator depends on its structure, materials, actuation mode and clamping (Sherrit et al. 1994).

Contraction and expansion of the active material in the lateral direction is resisted by passive material (cermet and metallic layers in Fig. 3) or material with a lower d31 coefficient. The structure is forced to bend under electrical excitation due to the stresses between these two layers. In Fig. 3 the result of this would be seen as flattening of the RAINBOW actuator when driven with an electric field parallel to poling. Bending of the actuator can produce significant axial displacement via amplification of the lateral displacement (a factor of over a hundred can be achieved). The amplification depends on the size and shape of the actuator, Young’s modulus and the material thicknesses of the passive and active materials. (Furman et al. 1994, Haertling 1997, Shih et al. 1997, Mossi et al. 1998, Wise 1998, Li et al. 1999) Optimal material thickness ratios for circular and rectangular unimorph type benders can be calculated through the maximum axial displacement capability of the actuator (Li et al. 1999)

24

( )2 6E E t t t t d E 1-L p np p np p np 31xhmax,circ 8 2 4 2 4 2 2E t E t 2E E t t 2t 2t 3t tp p np np p np p np p np p np

⎛ ⎞+ ν⎜ ⎟⎝ ⎠Δ = ×

⎛ ⎞+ + + +⎜ ⎟⎝ ⎠

(2)

( )2 2L L 6E E t t t t d E 1x y p np p np p np 31hmax, rect 8 2 4 2 4 2 2E t E t 2E E t t 2t 2t 3t tp p np np p np p np p np p np

⎛ ⎞ ⎛ ⎞+ + − ν⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠Δ = ⋅

⎛ ⎞+ + + +⎜ ⎟⎝ ⎠

(3)

where (tp) and (tnp) are thicknesses, (Ep) and (Enp) are Young’s modulus of the piezoelectric and non-piezoelectric material respectively, (ν) is Poisson’s ratio of both materials, (E) is electric field, (d31) is lateral piezoelectric coefficient, (Lx) is length or diameter and (Ly) is the width of the actuator. The analytical solution assumes that the unimorph is standing freely on the support, the highest displacement (Δhmax,circ and Δhmax,rect) occurs in the centre of the actuator and that the Poisson’s ratio is the same for the piezoelectric and non-piezoelectric materials. Fig. 7 shows theoretical and experimental results obtained by Li et al. (Li et al. 1999)

Fig. 7. Displacement of unimorph at a constant voltage as a function of the brass/PZT thickness ratios for squares and discs (Li et al. 1999).

The calculations and experiments in Fig. 7 were made for a PZT disc (∅25.4 mm and thickness 1.0 mm) and a PZT square (length 15.2 mm and thickness 0.381 mm) together with glued brass backing with different thicknesses (0.127-0.813 mm). Displacements can be increased by a factor of several times by adjusting the thickness of the materials. Using a ~0.38 thickness ratio of passive and active materials gives the maximum results for circular and square unimorphs as shown in Fig. 7. (Li et al. 1999) Even though the experiments and equations were for unimorph actuators, the same rules apply also for

25

pre-stressed actuators (Furman et al. 1994, Haertling 1997, Shih et al. 1997, Mossi et al. 1998, Wise 1998, Li et al. 1999, Wang & Cross 1999 b, Li et al. 2001 Schwartz et al. 2001).

2.3 Manufacturing methods and performance

Since the year 1994, several pre-stressing methods have been introduced. The earliest actuators utilized a mismatch of the coefficient of the thermal expansion (CTE) between different layers. Recently, sintering shrinkage has also been utilised as an agent of the pre-stress. Manufacturing methods and properties of pre-stressed benders and unimorph actuator are gathered into Table 1 and 2, used later in calculations and comparison.

Table 1. Manufacturing methods and properties of unimorph and pre-stressed benders.

Actuator Unimorph RAINBOW CERAMBOW CRESCENT RAINBOW type actuator

Pre-stress No pre-stress

CTE and volume change CTE CTE Sintering shrinkage

Method Gluing in room temp.

Reduction Soldering Thermally cured glue

Co-sintering

Piezo APC PZT-857

PZT 5A 1) PZT 5H 1) PLZT 1.0/53/47

PLZT 5.5/53/47 2)

PKI550 (PZT 5H)

PZT +5wt.% Ag2O

Sample D3 33 % metal

25 % metal

RB1053-8 - Sample in 250 °C

-

Lx x Ly [mm] ∅25.4 ∅25.4 ∅25.4 ∅22.4 ∅31.75 26x11 ∅31.0 tp [µm] 1000 255 381 322 432 1090 400 tnp [µm] 381

(brass) 126

(cermet) 127

(cermet) 110

(cermet) 127

(brass) 370

(SS302) 400 (PZT +

28wt.% Ag2O) E [MV/m] +0.40 ±0.75 ±0.75 +1.60 +1.04 0.185 ±0.325 F [N] - 0 0 - 0 - - f [Hz] - 1 1 - - 100 0.15 Ep [GPa] 62.5 69 1) 72 1) 74.2 74.2 3) - - Enp [GPa] 110 4) 62.6 3) 62.6 3) 62.6 110 4) 210 4) - d31 [pm/V] -295 -171 1) -274 1) -271 -271 3) - - νp 0.3 5) 0.35 1) 0.34 1) 0.374 0.374 3) 0.3 - νnp 0.3 0.342 3) 0.342 3) 0.342 0.34 0.3 - Δhmeas [µm] ~6.9 ~178 ~135 ~117 ~88 ~25.5 6) 9 References (Li et al.

1999) (Wise 1998)

(Wise 1998)

(Li et al. 1997)

(Barron et al. 1996)

(Chandranet al. 1997)

(Pearce & Button 1998)

Length or diameter (Lx), width (Ly), thicknesses of piezoceramic (tp) and passive (tnp) material, electric field (E), force in measurement (F), frequency (f), Young’s modulus of piezoceramic (Ep) and passive (Enp) material, piezoelectric coefficient (d31) Poisson’s ratios of piezoceramic (νp) and passive (νnp) materials and measured displacement (Δhmeas). 1) Properties not mentioned in publication, properties of PZT 5H and 5A piezoelectric ceramics used in PRESTO experiments were used. 2) Note that composition used in experiments by Barron et al.

26

was PLZT (5.5/56/44) with slightly different characteristics. 3) (Li et al. 1997). 4) (Goodfellow 2005). 5) Not mentioned in publication, selected for analysis. 6) Cantilever clamped from one end.

Table 2. Manufacturing methods and properties of pre-stressed benders.

Actuator THUNDER Prindrex / d31 gradient actuator LIPCA Co-extruded PZN-PZT

Pre-stress CTE Sintering shrinkage CTE CTE Method Thermally cured

LaRC-SI™ Co-pressing and co-sintering Thermally cured

epoxy Co-extrusion and

co-sintering Piezo PZT 5A

1) PZT 5H

1) PZT

E70/E63 PZT

E76/ZnO 3195 HD (PZT 5H)

PZN-PZT

Sample 33% metal

25% metal

Thickness ratio 1

E76/ZnO + 4wt.% Sb2O3

LIPCA Model C1

-

Lx x Ly [mm] ∅25.4 ∅25.4 ∅25.4 ∅25.2 72.42x23 2) 50x20 tp [µm] 255 381 500 (PZT E70) 670 250 420 (PZN-PZT) tnp [µm] 126 (Al) 127 (Al) 500 (PZT E63) 330 280 3) 420 (PZN-PZT+

50wt.% Ag) E [MV/m] ±0.75 ±0.75 +1.00 +0.90 ±0.75 +1.43 F [N] 0 0 - - 0 - f [Hz] 1 1 0.05 0.05 1 - Ep [GPa] 69 1) 72 1) 63 (PZT E70) 62.5 67 74.5 Enp [GPa] 70.6 4) 70.6 4) 100 (PZT E63) 110 5) 80 d31 [pm/V] -171 1) -274 1) -180 (E70-E63) 6) -250 -190 -230 νp 0.35 1) 0.34 1) 0.3 0.3 0.31 0.3 7) νnp 0.345 4) 0.345 4) 0.3 0.3 0.31 0.3 7) Δhmeas [µm] ~138 ~114 16 40 ~1051 ~175 References (Wise

1998) (Wise 1998)

(Li et al. 2001)

(Li et al. 2001)

(Yoon et al. 2003)

(Yoon et al. 2005 a)

Length or diameter (Lx), width (Ly), thicknesses of piezoceramic (tp) and passive (tnp) material, electric field (E), force in measurement (F), frequency (f), Young’s modulus of piezoceramic (Ep) and passive (Enp) material, piezoelectric coefficient (d31) Poisson’s ratios of piezoceramic (νp) and passive (νnp) materials and displacement (Δhmeas). 1) Properties not mentioned in publication, properties of PZT 5H and 5A piezoelectric ceramics used in PRESTO experiments were used. 2) Total length 100 mm and total width 24 mm. 3) Bottom layer: glass/epoxy (90 µm) + carbon/epoxy (100 µm) + glass/epoxy (90 µm), top layer: glass/epoxy (90 µm). 4) (Goodfellow 2005). 5) Young’s modulus of carbon/epoxy 231.2 GPa in axis 1, and 7.2 GPa in axis 2 and glass/epoxy 21.7 GPa. 6) PZT E70 d31: -230 ± 30 pm/V, PZT E63 d31:-30 ± 10 pm/V. 7) Not mentioned in publication, selected for analysis.

The RAINBOW actuator is manufactured by a reduction process at elevated temperature (600-1000 °C) where one side of the piezoelectric ceramic disc is placed in a reducing medium. Typically, carbon is used as a reducing agent which reduces oxides of the piezoelectric material and produces layers of cermet and metallic lead on the disc. The different thermal expansion coefficients of the unreduced ceramic and the cermet and/or metal layers introduce asymmetric internal stresses during cooling. (Furman et al. 1994, Wang & Cross 1999 a) Additionally, reduction of the volume in the reduced layer induces

27

more stress into the actuator (Haertling 1994 b, Furman et al. 1994). High displacements of the actuators are obtained due to the contraction and expansion produced via the d31 coefficient, as explained in Chapter 2.2. RAINBOW actuators have generally exhibited the highest displacement capabilities among the pre-stressed bending actuators, excluding the recently developed LIPCA (Barron et al. 1996, Pearce & Button 1998, Wise 1998, Li et al. 2001).

CERAMBOW actuators are manufactured by soldering a piezoceramic onto a metal e.g. brass. Soldering was carried out at ~250 ºC on a hot plate where the solder was uniformly applied on a brass sheet. After heating the solder and metal sheet, the pre-heated piezoelectric disc was placed on the solder with applied weight. Pressure was applied during cooling for a short period of time in order to keep the ceramic and metal surfaces attached. When cooled, the CERAMBOW actuator bends due to the different CTE of metals and ceramic. The authors imply that stress gradients in the ceramic layer enable greater displacements than those produced by an average unimorph actuator. (Barron et al. 1996)

The CRESCENT actuator is also bonded with metal, but using a thermally cured adhesive, for example, epoxy which is cured at 200-400 ºC for 0.5–6 hours. CRESCENT actuators become domed and pre-stressed during rapid cooling in air from the bonding temperature. In optimum conditions at 250 ºC, CRESCENT actuators displayed superior characteristics compared to corresponding unimorph actuators. The reason for this was suggested to be due to an increased d31 piezoelectric coefficient that may be a consequence of specific domain structures formed during poling. On the other hand, it was indicated that using excessive mechanical pre-stress (a temperature of 350 ºC) decreased the displacement of CRESCENT actuators, resulting in inferior characteristics compared to unimorph actuators. (Chandran et al. 1997)

The THUNDER actuator (Fig. 8) differs from earlier ones in the sense that it typically utilises pre-stressing metal layers of both sides of the ceramic. This is especially beneficial because metals make the actuator very rugged for handling. The asymmetric stress state in the THUNDER actuator is accomplished using different thicknesses and/or different metals that are bonded at 300-325 ˚C temperatures. (Mossi et al. 1998, Wise 1998) Typically, thin steel and/or aluminium foils are bonded using sprayed LaRC-SI™ polyimide with N-methyl-pyrolidinone solvent (Mossi et al. 1998, Wise 1998). The actuator is curved during cooling due to differences in the CTE and strength of the materials. Recently, an improved version of THUNDER, called LIPCA, has been introduced. In this case the metals of the THUNDER actuator are replaced with composite materials in order to improve displacement capabilities and lighten the structure (Fig. 8). Bonding is realised with co-curing of a stacked carbon/epoxy, PZT wafer and glass/epoxy laminates. (Yoon et al. 2003, Haris et al. 2004, Kim et al. 2005) The LIPCA actuator has shown approximately three times higher free displacement compared to the THUNDER actuator, probably having the highest displacement capabilities among the pre-stressed benders (Mossi et al. 2003).

28

Fig. 8. THUNDER and LIPCA constructions (edited from Mossi et al. 2003, Yoon et al. 2003).

Bonding of bimorph or unimorph is typically carried out by physical bonding, for example glue, that can make the interface fragile or weak resulting in de-lamination, poor fatigue resistance, reduced performance and increased hysteresis. Strong and reliable bonding is required between layers which implies that the bonding layer is uniform and thin and the surfaces are clean. The interlayer is usually softer than the active or passive layers and this degrades performance to a certain degree. The interlayer should also be able to withstand higher temperatures, aging and stresses, especially in resonance devices. (Zhu & Meng 1995, Kugel et al. 1997, Uchino 1997, Wang & Cross 1999 b, Li et al. 2003, Schulz et al. 2003, Taya et al. 2003, Hall et al. 2005) Although good results may be obtained with adhesives, general problems associated with the bonding have led to development of co-fired bimorphs, unimorphs or monomorphs and functionally graded structures using sintering phenomena as a bond between layers (Dogan 1996, Schulz et al. 2003, Wang & Cross 1999 b). In the case of the bimorphs and monomorphs, co-firing has improved reliability, blocking force, lifetime, stiffness and operational temperatures and has also decreased the voltage requirements (PI Ceramic 2005, Colla et al. 2000, Galassi et al. 2001). Using co-fired ceramic and functional gradients one can produce monolithic structures with reduced hysteresis and stresses, thus extending the lifetime of the actuators (Zhu & Meng 1995, Takagi et al. 2003, Taya et al. 2003, Hall et al. 2005). It is also suggested that introduction of graded microstructures can maintain displacement capability on the same level with the bimorph (Takagi et al. 2003). In some cases the co-firing process can also produce pre-stress for actuators, using the different sintering shrinkages of CTE mismatch of at least two materials.

In the method introduced by Button and Pearce silver oxide doped soft PZT material was used. Doping of the PZT had several purposes: a) to lower the sintering temperature of the PZT b) to provide a conductive Ag layer through reduction of Ag2O at high temperature c) to obtain different sintering shrinkages of two layers due to different amount and reduction of the silver oxide. Two layers of PZT sheets in green state were laminated together having 5 wt.% and 28 wt.% of Ag2O respectively and 2 % difference in linear shrinkage. Circular plates were stamped from laminated tapes and co-sintered at 1000 °C for 2 hours in a closed crucible and on zirconia sand. A bent and curved actuator structure was obtained which had a piezoelectric upper layer and a conductive and passive bottom layer. (Pearce & Button 1998)

Li et al. also utilised co-fired materials and the difference in sintering shrinkage. Piezoelectric PrinDrex actuators with two different layer configurations were manufactured. In the PZT/PZT structure, piezoelectric materials with different d31 coefficients were co-pressed and co-sintered. These structures produced an inner-type

29

dome structure where the piezoelectric material with the larger d31 coefficient formed the bottom layer and the material with the lower d31 was the upper layer. The structure was initially domed in such a way that the layer with the lower d31 was on the convex side of the actuator. Since the material with the higher d31 coefficient had larger longitudinal shrinkage, axial displacement of the inner type actuator was upwards (i.e. the radius of the actuator decreased) under an electric field applied parallel to the polarisation. The highest displacement was obtained when the thicknesses of the layers were same. The actuator configuration with PZT/ZnO layers was flat, inner or outer-type domed depending on the thickness ratio and Sb2O3 content of the ZnO layer. An outer-type actuator is typical for pre-stressed bending actuators i.e. the piezoelectric material is the upper layer, subjected to tensile stress and the actuator is moving downwards when an electric field is applied parallel with the polarisation. The PZT/ZnO actuator produced higher displacements than the PZT/PZT actuator. The ZnO layer showed typical varistor behaviour under electrical testing and exhibited ~108 times higher conductivity than the PZT layer, hence the ZnO layer can be treated as an electrical conductor for the PZT layer. (Li et al. 2001)

The most recently developed pre-stressed bending actuator was introduced by Yoon et al. It was realised via the co-extrusion of PZN-PZT and PZN-PZT/Ag compositions and subsequent co-firing. Two layer and multilayer structures were fabricated using PZN-PZT as an active material and PZN-PZT/Ag as a passive material and electrode. Extruded structures were sintered at 900 °C for 4 hours in a sealed alumina crucible in a PbO atmosphere. Due to the thermal expansion difference of the layers the actuator became outer-type domed during cooling. A silver content increased the CTE of the PZN-PZT composition producing residual stress in the structure. In the case of the two layer (i.e. unimorph) actuator solution, larger concentration than 50 wt.% of silver caused too high stresses introducing cracks into the interface of the layers. Additionally, in order to achieve good electrical conductivity the concentration of Ag need to be optimised according to the thickness of the layers. (Yoon et al. 2005, Yoon et al. 2006 )

The RAINBOW actuator has a unique combination of high displacement and monolithic structure. Advantages derived from the structure are high reliability and fatigue resistance, compared to conventional piezoelectric bending actuators, due to the strong chemical bond between different layers. (Haertling 1994 b, Barron et al. 1996, Dausch 1997, Li et al. 1997, Wang & Cross 1999) However, commercial production of RAINBOW has ended and THUNDER is currently the only commercially available pre-stressed bending actuator (Niezrecki et al. 2001). The reason for this might be manufacturing issues. On the other hand, recently developed actuators utilise more feasible manufacturing processes but to date have shown lower displacement properties or non-monolithic structure.

In an optimal case, a high tensile stress is created in the piezoelectric material of the pre-stressed unimorph-type actuator enhancing the effective d31 coefficient, which describes the average electromechanical response of the piezoelectric layer (Li et al. 1997, Schwartz et al. 2001, Schwartz & Moon 2001). Higher stresses of the RAINBOW actuator have been suggested to be the reason for its higher displacements compared to THUNDER and CERAMBOW actuators (Wise 1998). Also non-uniform distribution of the internal stress through the thickness contributes to the field-induced displacement (Li et al. 1997). The mechanism contributing to the increased effective d31 coefficient is an

30

altered initial domain configuration with minimal suppression of the 90° switching by internal stress. A higher a-domain (domains parallel to the actuator surface) population compared to the corresponding bulk actuator was obtained for PLZT (lead lanthanum zirconate titanate) RAINBOW actuators. The initial a-domain population depends on the tensile stress of the actuator surface i.e. higher stress introduces more a-domains. (Li et al. 1997, Schwartz & Moon 2001) Especially important is the a-domain population after poling enabling 90° domain switching and greatly affecting the field-induced displacement. Although, only a fraction (~7 %) of the a-domains experienced 90° switching, significant enhancement in the d31 coefficient has been obtained due to the large anisotropy in switching. Calculations predicted d31=-1400 pm/V under ~50 MPa tensile stress for the surface of the RAINBOW actuators while measured values of the bulk actuators were approximately -190 pm/V. (Schwartz & Moon 2001) Similar results were also obtained by Li et al. for bulk actuators under transverse tensile stress (Li et al. 2002). The enhancing effect of the altered domain configuration is assumed also to be present in the other pre-stressed piezoelectric actuators. However, the requirements of the stress controllable domain structure have to be fulfilled in order to have a significant increase in strain by domain reorientation (Li et al. 1997).

In Table 1 and 2 actuator configurations with the highest displacement capabilities were selected and results obtained with support of both ends or the perimeter of actuators. Accurate comparison of different pre-stressed actuators is difficult and not possible in all cases due to the large number of different actuators, measurement configurations and unknown material parameters. However, a rough comparison is made in Table 3 using equations (2) and (3) together with displacements measured in publications. Since the shape of the actuators (rectangular, square or circular) has a significant impact on displacement capabilities both equations are required to extract the effective d31 from displacement data. The measured displacement is compared to the calculated displacement using actuator parameters of the publications. This value gives some estimation about the possible influence of pre-stress. The difference is assumed to be the product of the pre-stress and hence an altered part of the d31 piezoelectric coefficient. The effective piezoelectric coefficient d31 used in calculations to reach the measured displacement is also indicated in Table 3. The estimation method used is basically the same as that used by Schwartz et al., but for free-standing actuators (Schwartz et al. 2001). It should be noted, however, that the piezoelectric coefficient d31 changes also under high electric fields even though stresses are not introduced (Wang et al. 1999, Schwartz et al. 2001). In order to compare the actuators later with PRESTO actuators the equations were used to calculate the displacement of the actuator with the same thickness ratio of materials but scaled in size and shape (∅25 mm, piezoceramic thickness 0.25 mm). With the same size and thickness of the piezoelectric material the displacement at the optimal passive material thickness was also calculated. For both results the material parameters in Table 1 and 2 and the effective d31 value under ±0.75 MV/m electric field and zero force were used. However, the piezoelectric material was kept the same which has a significant impact on the displacement in terms of piezoelectric coefficients under pre-stress.

The thickness and Young’s modulus ratios of the passive and active materials have great importance in actuator displacement and load bearing capabilities. Therefore, it is important to notice that this approximation gives only a general perspective to compare

31

the magnitude of displacement of the different actuators because not all parameters are known accurately. Additionally, the effect of the interlayer is ignored in the unimorph model, but different Poisson’s ratios of the materials are taken into account in presented cases. (Li et al. 1999) In calculations of RAINBOW actuators, passive and active material thicknesses are assumed to be homogenous. In the case of the THUNDER actuators, the top layer is ignored and the calculation is made using only the properties of the bottom layer. In the case of the gradient actuator the d31 coefficient was obtained by reducing lower value of d31 from the higher value of d31 as they work in the same phase (Li et al. 2001). Due to the complexity of the composite structure of the LIPCA actuator it was not included in the calculations. Applied forces in measurements were ignored in all cases.

Table 3. Calculated performance for unimorph and pre-stressed benders according to properties listed in Table 1 and 2.

Actuator, material Δhmeas [μm]

Δhmax,circ

[μm] (Δhmeas/Δhmax,circ)-1

[%] d31eff

[pm/V] Δhmax,circ

1) [μm]

Δhmax,circ 2)

[μm]

RAINBOW PZT 5A ~178 46 284 -656 176 176 PZT 5H ~135 46 192 -801 199 214 PLZT (1.0/53/47

RB1053-8) ~117 44 165 -718 176 188

THUNDER PZT 5A ~138 47 192 -499 136 136 PZT 5H ~114 48 138 -652 168 178

Prindrex / d31 gradient actuator

PZT E76 / ZnO + 4 wt.% Sb2O3

40

18

121

-554

175

178

PZT E70/E63 16 15 6 -192 47 61 CERAMBOW

PLZT 5.5/53/47 ~88 50 77 -480 136 140 Co-extruded PZN-PZT

PZN-PZT ~175 149 3) 17 3) -270 66 80 Unimorph, no pre-stress

APC PZT-857 ~6.9 6.5 6 -312 100 100 1) Scaled actuator size (piezoceramic ∅25x0.25 mm) calculated using d31eff. 2) Scaled actuator size (piezoceramic ∅25x0.25 mm) calculated using d31eff and optimal thickness ratio. 3) Δhmax, rect

According to Table 3, the RAINBOW actuators exhibited the highest effective d31 and more importantly the highest difference between measured and calculated values. Other pre-stressed actuators also show enhancement of the effective d31 coefficient. At the same time, unimorph actuator without pre-stress exhibited only a 6 % difference between the measured and calculated displacement. Due to the earlier mentioned assumptions and simplified cases in equations 2 and 3, results obtained here for the RAINBOW actuators

32

are higher than those suggested by Li et al. where the effective d31=-508 pm/V (for sample RB1053-8) value was obtained using a FEM analysis of the structures (Li et al. 1997). However, over-estimation of the effective d31 can be accepted as it is present in the values of every actuator. It should also be noticed that the differences between actuators are large, which makes comparison possible and clearly indicates the magnitude of the stress-enhanced effect.

3 Measurement methods and manufacturing of RAINBOW and thick film actuators and their properties

In order further to develop manufacturing methods of pre-stressed actuators, a deeper understanding is required about the existing methods. As mentioned earlier, present methods utilise either a CTE difference or sintering shrinkage. Since these are evidently very practical methods in the realisation of pre-stressed actuators they were first studied individually in the form of screen-printed thick film and RAINBOW actuators. RAINBOW represents the most desirable properties in its class while thick film actuators are exceptionally easy to produce. Both actuators have been well studied and information is widely available in literature thus enabling the synthesis of theory and practice. At the same time the knowledge basis was expanded about the material properties and measurement methods.

3.1 Measurement and analysis methods

The hysteresis loop, remanent polarisation and coercive field measurements were carried out with a Radiant RTV6000HVS system (Radiant technologies, USA). Different temperatures were applied using a controlled heating plate and an external thermometer which was placed near the piezoelectric sample in silicone oil. In order to keep a homogenous temperature in the oil it was mixed with a magnetic stirrer and the temperature was allowed to stabilise before measurements. Temperature accuracy was about ±1 °C. The same setup was used in poling while a sufficient DC voltage level was applied using the amplifier of the Radiant RTV6000HVS system and an HP33120A signal generator. The fired thickness of AgPd, LTCC and PLZT and curvature of the samples was measured using a profilometer (DEKTAK3ST Surface Profiler) for small dome heights and a micrometer screw (Mitutoyo) for larger dome heights. Measurements of the material thickness in the cross-section of the RAINBOW actuators were carried out with a Leitz Ergolux optical microscope having ~2.75 µm spacing between grid lines at the selected magnification. Cross-section pictures were taken with an Olympus SZ-60 optical microscope equipped with a digital camera system and a Micrion 2500 FIB

34

(Focus Ion Beam) system. Elementary analysis was carried out by a Scanning Electron Microscope (SEM) Jeol JSM-6400 equipped with an Inca Energy Dispersive Spectrometer (EDS).

Displacement measurements were carried out by a system based on a Michelson interferometer (Moilanen & Leppävuori 1993, Moilanen & Leppävuori 2001, Paper III). A schematic drawing of the measurement system is presented in Fig. 9.

Fig. 9. Schematics of the measurement system (Paper III).

The laser beam of the interferometer is reflected back from a silicon mirror to the detector of the interferometer. Due to the displacement of the mirror, the reflected beam experiences continuous phase shift compared to the reference beam of the interferometer. Therefore, the reflected and reference beams have continuous change in interference on interference plane and as a consequence, intensity variations can be detected. In order to detect displacements less than a wavelength of the beam, the reference beam is divided into two parts with 90° phase shift. The two interference signals of the beams in 90° phase shift are detected with separate detectors, converted to voltage signals and lead to a digital oscilloscope. (Moilanen & Leppävuori 1993) From the oscilloscope, the captured data is taken to a PC via GPIB for analysis using software based on LabVIEW. The program transforms the two voltage signals to a Lissajous pattern and calculates the total displacement based on the fact that one full cycle of the Lissajous pattern is half of the wavelength, ~390 nm. The same program also feeds a driving signal for the actuator through a signal generator and amplifier (Radiant technologies, USA) via GPIB. The resolution of the Michelson interferometer system is ~2 nm and the noise level of the

35

system is approximately 5 nm, depending greatly on disturbances of the surroundings. (Moilanen & Leppävuori 2001)

Loads were applied to the RAINBOW actuators by different masses (0.3-1.5 N) with a Round Robin jig (Moilanen & Leppävuori 2001). Masses were applied to the system by a thread in order to keep them steady. The weight of the masses was measured with a precision scale and accuracy of the weights was ~0.1 g. Later on the system shown in Fig. 9 was developed which utilised a micrometer screw (Newport, USA) and springs for loading (0.3-3.0 N) the PRESTO actuators (Paper III-IV). In this case the applied force was read out from the force sensor (KYOWA LUB-20KB) via an instrumentation amplifier (WGA-650A KYOWA). The accuracy of the force measurement system was about 0.01 N, but in practice forces were applied with 0.05 N accuracy. In both cases loads were applied to the centre of the actuators with a steel ball in order to measure the axial displacement from that point. Actuators were aligned on the jig using a circular frame having a diameter 0.5 mm greater than actuators. The measurement frequency of 10 Hz was well below the resonance of the spring (resonance of the free spring ~120 Hz) and force sensor (~500 Hz). The measurement frequency was also well below the bending resonance of the actuators. The mechanical part of the measurement system was placed on an air cushioned optical table (Newport, USA) in order to minimise mechanically transferred disturbances from the surroundings.

In the case of the thick film actuators, a cantilever configuration was used in the measurements. Cantilevers were supported from one end and the tip displacement as a function of frequency and voltage was measured with the interferometer using small glued mirrors. Firstly, a mirror was glued to reflect the beam back in the thickness direction from the end of the piezoelectric material while the other end was fixed with glue from the point where the active material ends. Secondly, a silicon mirror was glued in the same place, but the reflecting direction was along the longitudinal axis of the actuator. Since the initial geometry of the actuator was known, the d31 piezoelectric coefficient could be determined from the changed geometry during excitation in quasistatic conditions (10 Hz). (Paper II)

The displacements of the PRESTO actuators with a glued steel layer and external amplification mechanism were measured with the interferometer system by gluing a mirror on the centre of the actuators. During the measurement actuators were only subjected to the load applied by the mechanism as they were attached to the piezoelectric actuators. (Paper V)

3.2 RAINBOW actuators

In the experiments, discs of soft piezoelectric ceramics PZT 5A and PZT 5H of Morgan Electro Ceramics were used (Juuti et al. 2001, Juuti et al. 2004, Paper I). These materials provide high d31 coefficients and are used for pre-stressed actuators (Wise 1998, Wang & Cross 1999, Berlincourt et al. 2005). The diameter of the discs was 25 mm and the thickness 500 µm and 250 µm. Different processing temperatures and reduction times were used to study the reduction process and its effect on the RAINBOW actuators. The processing temperature has an effect on the reduction rate and composition of the reduced

36

layer. The aim was to achieve a sharp and thin interface between the piezoelectric ceramic and the reduced ceramic (Fig. 10). (Paper I) Process conditions were also adjusted to produce high pre-stresses, i.e. dome heights, by maximizing the amount of metallic lead in the reduced layer.

Fig. 10. Cross-sectional view of the PZT 5A RAINBOW actuator reduced at 750 °C. a) photograph from edge of disc b) FIB image from centre of disc, conductive material can be seen to be brighter (Paper I). Pores in the reduced layer can be seen as black regions.

Processing was obtained by using tightly packed ZrO2 powder and a solid graphite block. The protective powder was applied over the piezoceramic disc and then formed with a cylinder (∅ 35 mm) and piston using hand pressure. The hot zone of a furnace (DEK) was set to the reduction temperature and the assembly was driven into the zone using a belt. After the reduction period the sample was driven out from the furnace. Usually the most suitable temperature was the lowest one, thus a long process was used achieving the required features and the best controllability over the reduction depth. However, the wear of the graphite in the long process was also a limiting factor, resulting, for example, in failure of the protection. Therefore, a higher temperature and faster process was used when high reduction ratios were aimed at. Reduction temperatures, time and the depth of the PZT 5A and PZT 5H materials are presented in Table 4. (Juuti et al. 2001)

Table 4. Reduction depths in the centre of actuators for different materials (Juuti et al. 2001).

PZT 5A PZT 5H Material 750 °C 800 °C 800 °C 850 °C

Depth [μm] 104 151 220 67 146 70 123 173 83 132 220 Time [min] 90 120 150 30 60 30 60 90 20 30 60

The reduction depth was approximately proportional to time as obtained for PLZT (Heartling 1994 c, Elissalde et al. 1996). However, reduction followed a parabolic law in results obtained by Wang and Cross for PZT 5H RAINBOWs (Motorola PZT 3203HD), hence the reduction depth was a function of the square root of time (Wang & Cross 1999

37

a). The differences are due to porous and inhomogenous reduced ceramic (Fig. 10) and therefore gaseous species (such as CO and CO2) can pass through the product layer making the process as reaction limited. Additionally, different compositions of the piezoelectric material have an effect on the reduction product and reduction ratio as shown in Table 4. (Wang & Cross 1999 a, Dausch 1996)

Generally, a shorter reduction time was obtained for PZT 5H type material (Table 4) than that obtained by Wang and Cross even though significantly lower temperatures were used. The reason is assumed to be due to differences in the protection. Protection has an effect on the reduction atmosphere as it controls the diffusion of the oxygen from the air. Therefore it is possible that in our case the portion of the oxygen coming from the PZT ceramic is higher, resulting in different reduction rates. Similarly, faster reduction rate can be a consequence of easier access of oxygen to the reaction. (Wang & Cross 1999 a) Overall, the results are well in line with those obtained by others (Haertling 1994 b, Elissalde et al. 1996, Wang & Cross 1999 a).

Typically, 500 µm thick samples were dome shaped (Fig. 11) and 250 µm thick samples had a saddle or mixed shape (between saddle and dome shape). Samples also had a flat and only slightly reduced region on the edge of the discs as shown in Fig. 11 and more closely in Fig. 10 (~0.5-1.0 mm perimeter from the edge).

Fig. 11. Cross-section of the 500 µm thick PZT 5A actuator with dome shape.

The cooling rate of the RAINBOW actuator is crucial to the manufacturing process because slow cooling results in re-oxidation and an insulating oxide layer. On the other hand, too rapid cooling can break the sample. Because the actuators were protected from thermal shocks by the ZrO2 powder and graphite block they were quickly removed from the processing temperature to room temperature using a conveyor belt.

After the process, samples were lightly ground to reveal metallic lead under the re-oxidised layer from the bottom of the disc. Electrodes were made on the whole area and both surfaces of the actuator using conductive silver paint (Electrolube) so that also a slightly reduced area on the edges was under electric field (Fig. 10 and 11). The silver paint had poor adhesion but it ensured good electrical connection and did not cause constraint of the actuators. Poling was carried out at room temperature for 30 min under electric fields of 2.5 MV/m and 4.0 MV/m for PZT 5H and PZT 5A, respectively. During poling the dome height of the actuators decreased to some extend i.e. the radius was increased. (Juuti et al. 2004, Paper I)

Displacements of the RAINBOW actuators were measured 24 h after poling. The measured PZT 5A and PZT 5H actuators (∅ 25 mm, thickness 500 µm) were reduced at temperatures of 750 °C and 800 °C, respectively. Displacement of a disc shaped PZT 5H

38

RAINBOW actuator as a function of electric field and under 1.5 N point load is shown in Fig. 12. (Juuti et al. 2004, Paper I)

Fig. 12. Displacement of the 500 µm thick (∅ 25 mm) PZT 5H RAINBOW and bulk actuators as a function of electric field (Juuti et al. 2004). Actuators were driven at 10 Hz frequency with 0V offset and under 1.5 N point load.

An initially disc shape PZT 5A RAINBOW actuators was cut from two sides with a Nd-YAG laser in order to study their electromechanical and geometric properties and any disruptive effect due to the laser cutting. After each cut, the relative permittivity, the profile of the cross-section and the displacement as a function of load were measured 24 h later. Measurements are explained in greater detail in Paper I. Cutting the actuators into a smaller size naturally reduced their load bearing capabilities, yet they showed reasonable properties even with large relative size reductions (Fig. 13). Laser cutting also had a small decreasing effect on the relative permittivity of the actuators due to the localised heating and disrupting effect of the beam. These effects could largely be avoided by poling the actuator after cutting. A larger decrease was observed under high electric field driving conditions due to the de-poling of the samples. (Paper I)

39

Fig. 13. Displacements of the 500 µm thick PZT 5A RAINBOW actuators as a function of load, driving field ±1.125 MV/m, frequency 10 Hz (Paper I). Actuators were poled under 4.0 MV/m electric field at room temperature for 30 min.

Both RAINBOW actuators exhibited a large bending motion several tens of micrometres in magnitude and sustained a high displacement with moderate loads (Fig. 12 and 13) (Paper I, Juuti et al. 2004). Properties of the RAINBOW actuators are gathered into Table 5. According to the comparison method presented in Chapter 2.3, the effective d31 is -191 pm/V for PZT 5A and -316 pm/V for PZT 5H disc shaped actuators, as shown in Table 6. In further calculations, the effective d31 value and parameters of Table 5 were used and a displacement of ~67 µm was obtained for different slices. The effect of the slicing of the RAINBOW actuators did not follow the equation (3) and measured displacement was decreased (Fig. 13). There are several possible reasons for these discrepancies. Firstly, cutting of the actuators decreased their stiffness and load bearing capability, which was not taken into account in the calculations. Secondly, narrower actuators were subjected to more load cycles which might have de-poled the actuators and decreased the d31 values. Finally, the differences might be due to a not entirely rectangular shape (the ends of the actuators were radiused) and the simplified shape of the passive layer in the calculations. Behaviour of the FEM (Finite Element Method) model of the actuators showed better correspondence than theoretical calculations because actuator geometry in the model was more precise. However, displacements were lower because material parameters supplied by the manufacturer were used. (Paper I)

40

Table 5. Material and driving parameters used in calculation of RAINBOW actuators.

Actuator RAINBOW RAINBOW Pre-stressing method CTE and volume change CTE and volume change Manufacturing method reduction reduction Piezoceramic PZT 5H PZT 5A Lx [mm] ∅25 ∅25 tp [µm] 320 270 tnp [µm] 170 (cermet) 210 (cermet) E [MV/m] ±0.75 ±1.125 F [N] 1.5 0.3 f [Hz] 10 10 Ep [GPa] 72 1) 69 1) Enp [GPa] 62.6 2) 62.6 2) d31 [pm/V] -274 1) -171 1 νp 0.34 1) 0.35 1) νnp 0.342 2) 0.342 2) Δhmeas [µm] 66 67 References (Juuti et al. 2004) (Paper I) Diameter (Lx), thicknesses of piezoceramic (tp) and passive (tnp) material, electric field (E), force in measurement (F), frequency (f), Young’s modulus of piezoceramic (Ep) and passive (Enp) material, piezoelectric coefficient (d31), Poisson’s ratios of piezoceramic (νp) and passive (νnp) materials and measured displacement (Δhmeas). 1) (Berlincourt et al. 2005, Morgan Electro Ceramics 2005, Sinocera 2005). 2) (Li et al. 1997).

Table 6. Measured displacements of RAINBOW actuators and their calculated performance.

Actuator, material

Δhmeas [μm]

Δhmax,circ

[μm] (Δhmeas/Δhmax,circ)-1

[%] d31eff

[pm/V] Δhmax,circ

1)

[μm] Δhmax,circ

2) [μm]

RAINBOW PZT 5H 66 57 15 -316 82 84 PZT 5A 67 60 11 -191 48 51

1) Scaled actuator size (piezoceramic ∅25x0.25 mm) calculated using d31eff and initial thickness ratio – see Chapter 2.3.

2) Scaled actuator size (piezoceramic ∅25x0.25 mm) calculated using d31eff and optimal thickness ratio.

Comparison of RAINBOW actuators shows ~2.0-2.7 times lower measured displacements for actuators produced here (Table 3 and 6). Subsequently, the actuators exhibited a 2.5 and 3.4 times lower effective d31 coefficient for PZT 5H and 5A, respectively. This was due to the non-optimised process in terms of composition and porosity of the reduced passive layer and the level of the pre-stress. It should be noted that the mechanical characteristics of the reduced layer in Table 5 are from results obtained by Li et al. (Li et al. 1997) It is expected that the actual Young’s modulus is lower. However, the introduction of 1/3 of Young’s modulus used in the calculation

41

would generate an effective d31=-443 pm/V, which is still lower than the value obtained by others.

The formation of the passive layer in the process has a large impact on the displacement characteristics of the actuator through pre-stressing and mechanical properties of the layer. Although RAINBOW actuators obtain high levels of pre-stress, bonding strength and effective d31 they are somewhat limited by the Young’s modulus of the passive layer. This is because the passive material is processed from the piezoceramic. In order to produce actuators with an improved performance (Table 3), the pre-stress level of RAINBOWs should be obtained by other methods (Li et al. 1997, Wise 1998, Li et al. 2001). Additionally, passive material with a higher Young’s modulus compared to RAINBOWs has to be realised with a high bonding strength. A more feasible manufacturing process is also crucial to further utilisation of the actuators.

3.3 Thick film actuators

Thick film actuators were manufactured from laboratory made PLZT (lead lanthanum zirconate titanate) powder by screen-printing. In order to reduce the sintering temperature of the PLZT paste, 4 wt.% of glass frit was used. PLZT actuators with a size of 30.5×9.0×0.09 mm were printed on an alumina (96 %) substrates which had AgPd electrodes on them. The top electrode was printed using the same AgPd paste. A detailed description of cantilever actuator, compositions and the manufacturing process is given in Paper II. Actuators were poled in a silicone oil bath under 5.0 MV/m at 100 ºC for 30 min. Over 24 h after poling, the profile of the actuators, d31 piezoelectric coefficient and tip displacement as a function of frequency with different driving voltages were measured (Fig. 14). (Paper II)

42

Fig. 14. Displacement of the tip of the cantilever as a function of frequency with 5 Vpp and 10 Vpp voltages (Paper II). First and second bending resonances appeared at 294 Hz and 1863 Hz frequencies, respectively.

The actuators exhibited a piezoelectric coefficient d31=-34.6 pm/V at 10 Hz frequency and under 5.0 MV/m electric field. This value can be compared to values determined by the earlier calculations (Table 3 and 6). It should be noted, however, that the d31 coefficient was obtained on the surface of the thick film actuator. The hysteresis curve was measured from poled and measured samples showing a remanent polarisation of 1.36 μC/cm2 at 100 °C under 5.0 MV/m electric field. These values are compared to values obtained from PRESTO actuators in Chapter 4.4. (Paper II)

The dome height at the centre of the cantilever was about 9 μm when supported from both ends during measurement (Paper II). The actuators showed an inner-type domed structure due to sintering shrinkage of the PLZT and AgPd paste and a CTE difference between alumina, PLZT and AgPd. This means that the free surface of the alumina was subjected to tensile stress while the PLZT was under compression. This is an unwanted state in terms of enhancing the piezoelectric actuation with the pre-stress because the compressive stress tends to decrease the effective |d31| coefficient (Li et al. 2001). Due to the solid substrate, sintering of the piezoelectric paste is constrained which can also lead to reduced piezoelectric characteristics, as proven by many others (Simon et al. 2001, Torah et al. 2004 a). Additionally, paste preparation and lowering the firing temperature have significant effects on the piezoelectric properties (Holc et al. 1999, Simon et al. 2001, Walter et al. 2002, Torah et al. 2004 b, Hrovat et al. 2006). The actuators exhibit a low performance compared to the actuators in Tables 3 and 6 and the advantages of the substrate with a high Young’s modulus and strong bond with sintering could not be utilised efficiently due to relatively low piezoelectric d31 coefficient.

In spite of the low piezoelectric characteristics of the thick film actuators in Paper II, they produced large displacements at resonance with low voltages (Fig. 14). This kind of

43

actuator can be utilised as a low force actuator, a resonant device and a sensor and the manufacturing method is feasible for mass-production (Aoyagi et al. 2002).

4 Pre-stressed electroactive actuator PRESTO

The trend in pre-stressed actuators is towards sintered monolithic structures with improved piezoelectric properties. Sintering gives several advantages such as a strong bond between layers, higher reliability, functional gradients, improved performance, good control of the thickness and miniaturisation. It also reduces costs and labour-intensive work phases such as cutting, polishing, and bonding. In the perspective of miniaturisation and mass-production, techniques such as screen-printing and tape casting are very attractive for utilisation in new actuators. In spite of good manufacturability, new actuators should also exhibit high piezoelectric coefficients initially and enhanced piezoelectric properties after the process utilising the high tensile stress in piezoceramics. In order to improve actuator characteristics even further, passive material should have a high Young’s modulus and a strong bond with the active material.

The idea of the PRESTO, the manufacture of different configurations and their electrical and electromechanical properties are presented and compared to corresponding RAINBOW and THUNDER actuators. Furthermore, effective d31 piezoelectric coefficients of the PRESTO actuators are calculated and compared to results obtained earlier from other pre-stressed actuators.

4.1 Idea and principles of the PRESTO

The basic idea of the PRESTO actuator developed in this thesis is that sintered piezoelectric ceramic is used as a substrate for the post-processed pre-stressing layer. As a result, good piezoelectric properties are obtained because sintering temperature is not reduced and shrinkage adjustments are not required as in the LTCC piezoelectric materials obtained by Corker et al. and Wolny (Corker et al. 2000, Wolny 2000). Additionally, in the different manufacturing processes piezoelectric, mechanical and dielectric properties can diminish greatly (Holc et al. 1999, Corker et al. 2000, Wolny 2000, Simon et al. 2001, Walter et al. 2002, Torah et al. 2004 b). Since high displacements and moderate load bearing capability are required from the new actuators it is particularly important that crucial properties, such as the d31 coefficient, are as high as possible. Consequently, it is suitable to use well optimised commercial bulk materials

45

obtaining high piezoelectric coefficients which might even be enhanced in the configuration of the PRESTO actuators.

A primary limitation is that the substrate material has to tolerate the processing conditions and the pre-stressing process. A pre-stressing material, on the other hand, is a material that can be sintered onto piezoelectric ceramic e.g. ceramic green sheet or metal conductor paste. In practice, materials sintered at temperatures below 1000 °C are desirable due to the lower cost and the avoidance of harmful effects for the piezoelectric ceramic. Piezoelectric properties can be reduced due, for example, to lead loss (lead oxide evaporates at around 800 ºC) and over sintering if temperatures near the sintering temperature of the piezoelectric material are used (Pokharel & Pandey 1999, Torah et al. 2004 b, Wang et al. 2004).The pre-stressing material should also exhibit a strong bond with the substrate material in order to introduce high stresses through sintering shrinkage and CTE mismatch. Adhesion, which is typically the weakest link, dictates the amount of pre-stress. Use of materials with a higher Young’s modulus should show an improvement over conventional actuators. Overall, an introduced pre-stressing process has an effect on displacement, load bearing capabilities and moment of inertia via stress, adhesion and properties of the pre-stressing layer.

The closest counterparts for PRESTO actuators are pre-stressed actuators produced by co-firing. In the co-firing, materials in the green state (i.e. not solid) are sintered in the same process while in the post-firing process some materials are already sintered (i.e. in solid state). Co-fired pre-stressed actuators utilise either sintering shrinkage or CTE mismatch for mechanical pre-stress. In PRESTO actuators both phenomena are utilized. (Pearce & Button 1998, Li et al. 2001, Yoon et al. 2005) In the co-firing process, the difference in sintering shrinkage is smaller than in actuators made by the post-firing method.

Actually, the post-processing technique of the PRESTO is used in ceramic hybrid circuit manufacturing. In the proposed method, however, the substrate material is active and stress created between the substrate and added material is a necessary feature. In post-processed ceramic circuit manufacturing stresses are minimised by CTE matching of the inks (Koch et al. 1998). Additionally screen-printing, used in the realisation of the PRESTO actuators, is a commonly used manufacturing method for the electrodes of the piezoelectric ceramics. Similarly, LTCC is widely used in the car industry, electronics manufacturing and in the telecommunication industry (Heyen et al. 2003, Jantunen et al. 2003).

Mass-production of actuators is possible by utilisation of the screen-printing and LTCC manufacturing technologies. In PRESTO manufacturing, existing production equipment can be used, special devices are not required and only very small changes are needed for standard processes e.g. changing of paste and printing thickness. Manufacturing processes are also well controllable and commercially available and optimised materials can be used for active and passive layers of the PRESTO actuators. Therefore, actuators with uniform quality can be expected with the minimum amount of labour-intensive work phases. In addition, work phases are minimised when conductive material is used as a pre-stressing layer. By this means electrodes and pre-stress are realised with a single process while in other methods electrodes and pre-stress are made in separate processes (Barron et al. 1996, Chandran et al. 1997, Pearce & Button 1998, Wise 1998, Li et al. 2001, Yoon et al. 2005).

46

4.2 Manufacturing of PRESTO actuators

PRESTO actuators were manufactured from 250 μm and 375 μm thick PZT 5A and PZT 5H discs of the Morgan Electro Ceramics. Pre-stressing of the actuators was carried out using AgPd thick film paste or LTCC tape. The actuator configurations used in Papers III-V are shown in Fig. 15. (Paper III) Different thicknesses of the AgPd paste (emulsion thickness 20-60 μm, referred in Figs. 17-19) was screen-printed obtaining fired thicknesses in the range of ~12-33 μm. The opening in the mask was 24.5 mm in diameter to prevent the paste from going over the edges of the sample. The samples were sintered at 850 °C in a belt furnace. In Paper III the same AgPd paste was used for the top electrodes (~12 µm) and the pre-stressing layer with total thickness of ~12-45 µm. Alternatively, the top electrode was made using conductive silver paint (Electrolube) after firing in order to study the constraining effect of the electrodes. The AgPd pre-stressing layer of the sintered sample consists of ~69 wt.% of silver and ~23 wt.% of palladium. (Paper III)

In the pre-stressed structures made with LTCC tape (DuPont 951, green thickness 114 µm, fired thickness ~96 μm), the top and bottom electrodes were first printed and fired on the piezoelectric discs using the AgPd paste (fired thickness ~12 μm) and the mask mentioned above. To make an electrical connection through the LTCC layer, a via with a diameter of 1.1 mm was laser-cut into the green tape. The green tape was cut into the same size as the piezoelectric disc and placed on one side of the disc with fired electrodes. The multilayer structure was installed into a vacuum pouch and laminated at 20 MPa and 70 °C for 10 min using a PTC IL4008 isostatic laminator. After firing at 875 °C in the belt furnace, the actuator domed, and the electrical connection was applied to the bottom surface using silver paint through the via hole. In Paper III actuators were poled at room temperature under 2.5 MV/m electric field for 30 min. After 24 h had elapsed after poling, the actuators were aged for 105 cycles under ±0.15 MV/m electric field, 10 Hz frequency and 0.3 N load. (Paper III)

Fig. 15. PRESTO actuators manufactured by using a) conductive and b) ceramic layer as a pre-stressing material (Paper III).

In Papers IV and V the pre-stressing was carried out with the same AgPd paste as in Paper III but the top electrode was replaced by Ag paste in Paper IV. Also, the same

47

masks were used for pre-stressing layer and top electrode. In Paper IV, after a 24 h period had elapsed from the firing, the hysteresis loop was measured using temperatures in a range of 25-125 °C and an electric field of 0.5-7.0 MV/m. The samples were poled with a 30 min poling time using the same temperatures and electric fields as in the hysteresis loop measurements and the displacement characteristics were measured 24 h after poling. Furthermore, the actuators were aged for 105 cycles before measuring displacement with an electric field of ±0.15 MV/m, 10 Hz and 0.3 N. (Papers IV-V)

In Paper V, external mechanical amplifiers utilising PRESTO actuators were manufactured. PZT 5H PRESTO actuators with thickness of 250 μm and 410 μm (∅25 mm) with 33 μm thick AgPd pre-stressing layers were poled at room temperature under a 5.0 MV/m electric field for 30 min. Steel layers with diameter of 35 mm and thickness of 25 μm for thinner and 50 μm for thicker actuators were glued after poling onto the pre-stressed side of the discs using cyanoacrylate glue. Electrical connection was achieved with a small dot of silver paint between steel and electrode which was applied in the centre of the disc just before gluing. The actuator installed on the base of the mechanical amplifier had 25 mm spacing to move as the steel around the disc was clamped tightly with screws and plastic plates. (Paper V)

4.3 Properties of the PRESTO actuators

Different configurations of the PRESTO actuators were tested in Paper III. The manufactured PRESTO actuators were of the outer-type i.e. the piezoelectric material was on the convex side of the disc. Doming of the structures for high and non-existing CTE mismatch with layer to be sintered (black layer) are presented in Fig. 16 a) and b), which correlate with the actuators pre-stressed with AgPd and ceramic layers.

Fig. 16. Curvature of actuator with different CTE mismatch.

The level of the pre-stress was roughly indicated by the dome height (material thicknesses are subtracted from the results) as shown in Table 7, which gathers the information of different actuator configurations. The first part of the configuration label (e.g. F3 in F3-1) refers to the figure number in Paper III while the second part gives an order in displacement performance at the lowest force. It should also be noted that the fired thickness of the AgPd top electrode is ~12 μm which is referred to as a constraining layer. In contrast, the Ag paint top electrode has an insignificant constraining effect. (Paper III)

48

PRESTO actuators exhibit dome heights comparable with the RAINBOW actuators (Li et al. 1997). The thickness of the active material reflected on the dome height, thus thicker material decreased the height in all actuators. In the cases of 250 µm thick actuators, higher dome heights were observed when the top electrode was applied using Ag paint after firing (Table 7). The dome heights of the actuators decreased significantly due to poling. This was expected as large stresses, opposite to the pre-stress, are generated in poling. The largest decrease of the dome heights was observed for the actuators with the highest dome heights as a consequence of their lowest initial stiffness which provides the spring force against poling-induced stresses. Based on this assumption, stiffer actuators with ceramic pre-stressing material should exhibit a smaller decrease in dome height. Surprisingly, these actuators produced about the same decrease in dome height as their closest counterparts with a metal pre-stressing layer. The reason for this is the same as in displacement generation during excitation. A thicker layer is stiffer, but it also has a better ratio of active and passive material generating higher displacements, as shown later (Li et al. 1999). Subsequently, some of this displacement will remain due both to residual stress from the poling and to decreases in asymmetric net pre-stress. (Paper III)

Table 7. Properties of PRESTO actuators (Paper III).

Material, label

Thickness of bottom layer [μm]

Top electrode material

Thickness of PZT [μm]

Height before poling [μm]

Height in test [μm]

Shape in test

PZT 5A F3-1 45 1) AgPd 375 435 215 dome F3-2 33 1) AgPd 375 470 300 dome F3-3 35 1) AgPd 375 335 270 dome F3-4 24 1) AgPd 375 320 260 dome

PZT 5A F4-1 33 1) AgPd 250 725 665 saddle F4-2 12 1) Ag paint 250 410 220 dome F4-3 24 1) AgPd 250 400 360 dome

PZT 5H F5-1 33 1) AgPd 250 470 345 mixed F5-2 33 1) Ag paint 250 620 380 mixed F5-3 33 1) AgPd 375 450 230 dome F5-4 33 1) Ag paint 375 270 140 dome

PZT 5H F6-1 96 2) AgPd 250 410 240 dome F6-2 96 2) AgPd 375 355 160 dome

PZT 5A F6-3 96 2) AgPd 250 450 260 dome F6-4 96 2) AgPd 375 410 230 dome

1) Bottom layer material: AgPd 2) Bottom layer material: LTCC DuPont 951

Same shapes found by Haertling for RAINBOW actuators could be seen for PRESTO actuators in Table 7. (Haertling 1997, Paper III) Poling changed the shape of actuators

49

with thicknesses of 250 µm from mixed to dome shape or from saddle to mixed shape, while 375 µm thick actuators were initially dome shaped. Actuators pre-stressed by the ceramic layer exhibited initially mixed or dome shape and after poling only dome shape due to their lower generated pre-stress and larger total thickness. (Paper III) The lower pre-stress and dome height obtained with the ceramic pre-stressing material is due to the similar coefficients of thermal expansion (CTE) of the LTCC and piezoelectric materials (PZT 5A ~2.5 ppm/K and DuPont 951 ~5.8 ppm/K, both near room temperature). (Berlincourt et al. 2005, CTS Microelectronics 2005) The actual pre-stressing is caused by the lateral sintering shrinkage of LTCC on the piezoelectric disc, although the 13 % shrinkage value of a freely standing LTCC DuPont 951 tape was not reached. (CTS Microelectronics 2005, Paper III)

It can be seen (Table 7), that even a thin layer (33 µm) of the metal produces dome heights comparable to or higher than a significantly thicker (96 µm) ceramic layer. This is due to a larger CTE mismatch between PZT and AgPd. The pre-stressing metal layer thickness also had a direct effect on dome heights of the actuators. An anomaly of this behaviour is observed in F3 cases in Table 7 due to different processing steps. In the case of F3-2, the top electrode and the pre-stressing layer were sintered at the same time. Conversely, in the cases labelled as F3-1, F3-3, and F3-4, the bottom and top electrodes were first printed and sintered, followed by another sequence of printing and sintering to form the pre-stressing layer. As a consequence of earlier fired layers, structure is stiffer and lower asymmetric pre-stress and dome height (Table 7 – see samples F3-1, F3-2 and F3-3) was obtained with the pre-stressing layer. Also, the electrode beneath the pre-stressing layer may soften and relief stresses produced via sintering shrinkage. (Paper III)

When different piezoceramics were compared, higher dome heights were observed for PZT 5A actuators under initial conditions and after poling. This might be due to a lower value of Young’s modulus. It should also be noted that differences in the CTE of PZT materials introduced slightly different stress levels in the actuators. (Hooker 1998) In general, different actuator configurations show significant differences and careful consideration has to be made in actuator design and manufacturing. (Paper III)

Displacements were measured under ±0.75 MV/m electric field at 10 Hz frequency using a 0.3-1.5 N point load for AgPd and 0.3-3 N point load for LTCC pre-stressed actuators. The results of PZT 5A PRESTO actuators with AgPd pre-stressing layer show that the displacements achieved were proportional to the thickness of the pre-stressing conductive layer (Fig. 17). (Paper III)

50

Fig. 17. Displacements of the 375 μm thick PZT 5A PRESTO actuators pre-stressed with AgPd layer as a function of force (Paper III).

Additionally, the results indicate an enhanced effective d31 coefficient especially in the case of the samples F3-2 and F3-3 (Fig. 17). These actuators have pre-stressing layer thicknesses of ~33 µm and ~35 µm, respectively, but due to the different processing steps, the pre-stress and displacement levels are different. (Paper III) The better performance of F3-2 is the result of higher pre-stress via co-firing of electrode and pre-stressing layer as concluded by others for different pre-stressed actuators (Li et al. 2001, Schwartz et al. 2001).

Fig. 18 presents the general results of Paper III, where thinner PZT 5H actuators generate higher displacements than thicker ones. This is also an obvious result that can be derived from equations (2) and (3). Additionally, the displacement capability of the thinner actuators depends more upon the load (Fig. 18 and 19 and Table 8). There was also a significant difference in load bearing capability of the different actuator materials, where the PZT 5A produced lower displacements but was more inert for loads than PZT 5H. The higher displacement of PZT 5H material can be understood from its initially larger d31 piezoelectric coefficient (PZT 5H: d31= -274 pm/V, PZT 5A: d31= -171 pm/V, manufacture’s information). (Berlincourt et al. 2005, Paper III)

51

Fig. 18. Displacements of the 250 μm and 375 μm thick PZT 5H PRESTO actuators pre-stressed with AgPd layer as a function of force with or without constraining layer (Paper III).

Displacement results for 250 µm thick PZT 5H actuators pre-stressed with AgPd in Fig. 18 show that the constraining top electrode has a significant reducing effect on displacement capabilities as a function of load. The actuators without a constraining layer generated more displacement at higher loads than their constrained counterparts. The same result was obtained also for 250 µm thick AgPd pre-stressed PZT 5A actuators in Paper III. Similarly, constrained 250 µm thick actuators obtained lower dome heights (Table 7), and hence lower tensile stresses. The constraining top electrode increases the stiffness of the actuator at the cost of the displacement and dome height. (Paper III)

The thicker PZT 5H actuators, however, showed only a slight decrease in their displacement as a function of load, regardless of whether or not a constraining electrode was used (F5-3 and F5-4). Such behavior under a load is natural due to the stiffer structure of the thicker actuators and the larger volume of the piezoelectric material. The shape of the actuator has also an effect on load bearing capabilities; the dome shape sustains higher loads than the saddle or mixed shape as obtained by Haertling for RAINBOW actuators (Haertling 1997, Paper III). The 375 µm thick PRESTO actuators were initially dome-shaped, which is a partial reason for their better load-bearing capabilities. Since only small differences between F5-3 and F5-4 actuators were observed it is assumed that the constraining effect is smaller in thicker actuators due the larger thickness. The relatively larger amount of material consequently contracts and expands more freely. Additionally, the net pre-stress level obtained with thicker actuators is smaller and therefore has no significantly different effects on different samples (F5-3 and F5-4). The 3.4 % difference can even be explained by variation of the piezoelectric coefficient in the bulk material.

Asymmetric net pre-stress is a result of stresses created by the top and bottom layers, which have opposite effects on the curvature of the actuator. The layer with the largest overall shrinkage in sintering and cooling determines the direction of curvature. In electrical excitation the direction of curvature is determined by the layer with the highest

52

stiffness, i.e. Young’s modulus and dimensions. Therefore, during excitation of the actuator with a painted silver electrode, the constraining effect and the opposing force are very small, and the structure is more free to bend into the direction determined by the pre-stressing layer. (Paper III)

Fig. 19. Displacement of the 250 μm and 375 μm thick PZT 5A and PZT 5H PRESTO actuators pre-stressed with LTCC tape as a function of force (Paper III).

Displacements as a function of load for actuators with a ceramic pre-stressing layer are presented in Fig. 19. The results showed that, in this case also, the displacements were larger when the actuator was 250 µm thick or the d31 coefficient of the piezoelectric material was higher (PZT 5H). The main difference between a AgPd and a ceramic pre-stressing layer was seen in the higher level of displacements at all forces. Improved displacement and load bearing capabilities are due to the fact that the used ceramic pre-stressing layer was thicker and the Young’s modulus of the ceramic material is higher than that of AgPd. Therefore, an increased Young’s modulus and thickness ratios of the passive and active materials were obtained which were decisive factors for the higher displacements. (Furman et al. 1994, Mossi et al. 1998, Wang & Cross 1999 b, Wang et al. 1999) On the other hand, the lower dome height (Table 7) indicates that the pre-stress level was lower (Schwartz et al. 2001, Li et al. 2001, Mossi et al. 2003). Thus an effect of the pre-stress in displacement could be seen more clearly from actuators pre-stressed by AgPd (Fig. 17 and Table 8).

Performance of the PRESTO, RAINBOW and THUNDER actuators with nearly the same diameter (∅ ~25 mm) and thickness were compared in Table 8 (Paper III).

53

Table 8. Average displacements of free RAINBOW and THUNDER actuators at 1 Hz frequency and PRESTO actuators at 10 Hz frequency under 0.3 N load and ±0.50 MV/m and ±0.75 MV/m electric fields. Relative change in displacement of the actuators under ± 0.50 MV/m driving field while loaded from 0.3 N to 1.5 N or 3.0 N (edited from Paper III).

Material, actuator, label Piezo/ pre-stress layer [μm]

Δhmeas under±0.50 MV/m

[μm]

Δhmeas under±0.75 MV/m

[μm]

Δhmeas change under ±0.50 MV/m

and load [%]

d31eff [pm/V],

label PZT 5A

PRESTO, LTCC Dupont 951, F6-3 250 / 96 40 72 +7 3) -216 PRESTO, AgPd/AgPd, F4-1 250 / 33 23 42 -54 2) -374 PRESTO, Ag paint/AgPd, F4-2 250 / 12 15 32 -19 2) -690 RAINBOW 1) 255 / 126 95 ~178 +1 3) -656 THUNDER 1) 255 / 126 75 ~138 +2 3) -499

PZT 5H PRESTO, LTCC Dupont 951, F6-2 375 / 96 49 89 -11 3) -413 PRESTO, AgPd/AgPd, F5-3 375 / 33 24 51 -9 2) -994 PRESTO, Ag paint/AgPd, F5-4 375 / 33 21 50 -7 2) -967 RAINBOW 1) 381 / 127 70 ~135 -11 3) -801 THUNDER 1) 381 / 127 60 ~114 -3 3) -652

1) (Wise 1998). 2) Applied force from 0.3 to 1.5 N. 3) Applied force from 0.3 to 3.0 N.

Load bearing capabilities of the 375 μm thick PZT 5H actuators are well in line with the results obtained by Wise, especially with RAINBOW actuators, while the displacement of the THUNDER actuator reduced less under load. Instead, the load bearing capability of 250 μm thick PZT 5A PRESTO actuators with AgPd pre-stressing layers was significantly lower than RAINBOW and THUNDER actuators. This was an obvious result of the thin pre-stressing AgPd layer. However, PZT 5A PRESTO actuators with ceramic pre-stressing layers showed better performance under load and had a 7 % increase in displacement when the load was increased from 0.3 N to 3.0 N. Displacement capabilities of the PZT 5H and PZT 5A PRESTO actuators were within a factor of 1.2-3.3 and 1.9-6.3 of corresponding RAINBOW and THUNDER actuators (Table 8), respectively. (Paper III)

Some of the differences in the values presented in Table 8 are due to different frequencies and loads used in measurements, which both tend to reduce the displacement of PRESTO actuators compared to RAINBOW and THUNDER actuators. (Paper III) A passive piezoelectric ring also reduces the obtained displacements as well as poling (Papers IV). The overall trend in the differences between the materials is the same, showing that the PZT 5A material also sustains loads better in the PRESTO structures (Table 8). Displacements of the PRESTO actuators, however, are smaller with PZT 5A materials, which is contradictory to the results obtained by Wise shown in Table 8 (Wise 1998). This can be due to the optimized structures and pre-stress levels in the THUNDER and RAINBOW actuators. For example under uniaxial compressive stress, the highest d31

54

and d33 coefficients are obtained at an optimal stress level as shown in Fig. 4 (Zhang & Zhao 1999, Paper III).

Accordingly, optimized structures and poling will improve the performance of the PRESTO actuators (Paper III-IV). Major improvements in the performance, however, can be expected due to optimization of the material’s Young’s modulus, pre-stressing level and the thickness of the pre-stressing layer enabling performances that are competitive with the results obtained by others for different structures (Furman et al. 1994, Mossi et al. 1998, Wise 1998, Wang & Cross 1999 b, Li et al. 2001). The displacements of THUNDER and RAINBOW actuators increased up to 205 % and 286 %, respectively, following optimization of pre-stress and the place of the neutral axis as suggested by Furman et al. and Mossi et al. (Furman et al. 1994, Mossi et al. 1998, Wise 1998). The same kind of improvements are anticipated for PRESTO actuators due to analogous behaviour of the actuator. Additionally, a competitive solution is obtained due to the straightforward manufacturing method, the integration possibility for LTCC and possibilities for high temperature operation. (Paper III) Optimisation, expected improvements and future possibilities are described in Chapter 4.4.

Electrical properties of PRESTO and bulk actuators under high electric field and at different temperatures were measured in Paper IV. Positive and negative remanent polarisation and coercive electric fields under highest electric field are presented at Table 9. (Paper IV)

Table 9. Hysteresis measurement results under the highest electric fields (Paper IV).

Sample Thickness [μm]

+Pr [μC/cm2] (25°C/125°C)

-Pr [μC/cm2] (25°C/125°C)

+Ec [MV/m] (25°C/125°C)

-Ec [MV/m] (25°C/125°C)

PZT 5A bulk 250 34.86/29.60 -35.61/-30.05 1.84/1.19 -1.67/-1.08 PZT 5A pre-stressed 250 25.77/27.72 -25.02/-27.42 2.87/2.38 -3.02/-2.31 PZT 5A bulk 375 34.63/29.22 -35.08/-30.42 1.73/1.17 -1.51/-0.98 PZT 5A pre-stressed 375 33.50/29.75 -32.90/-29.60 1.99/1.43 -2.23/-1.44 PZT 5H bulk 250 27.87/19.46 -29.07/-20.66 1.11/0.57 -1.05/-0.55 PZT 5H pre-stressed 250 21.33/20.21 -22.69/-20.36 2.04/1.23 -1.93/-1.17 PZT 5H bulk 375 29.30/21.18 -29.30/-21.33 1.16/0.54 -1.15/-0.52 PZT 5H pre-stressed 375 23.21/19.61 -22.16/-18.56 1.60/1.08 -1.77/-1.08

PRESTO actuators showed a significantly higher coercive electric field than their bulk counterparts, but a decreased remanent polarisation in Table 9. The highest remanent polarisation for the 375 μm thick PZT 5A bulk (34.63 μC/cm2) and PRESTO actuators (33.50 μC/cm2) were almost the same. (Paper IV)

In general the PZT 5H material had 9-37 % lower remanent polarisation and 20-54 % lower coercive electric field values than the PZT 5A material (Table 9). This result is well in line with the results reported earlier where a 31-33 % difference in remanent polarisation and 45-50 % difference in coercive electric field was measured by Dausch. (Dausch 1997, Paper IV)

The highest coercive field was achieved with 250 μm thick PZT 5A PRESTO actuators, 2.87 MV/m and -3.02 MV/m. ±Ec and ±Pr values for bulk and PRESTO actuators were higher than those obtained by Dausch, probably due to the higher electric

55

fields (Table 9). PZT 5A THUNDER actuators had higher remanent polarisation and about the same coercive electric field under ~2 MV/m (+Ec = 1.38 MV/m and -Ec = -1.4 MV/m) compared with PRESTO actuators. At higher electric fields remanent polarisations of the PRESTO actuators were closer to the values of THUNDER actuators and the coercive electric field was higher. It should also be noted that the thickness of the piezoelectric material in the THUNDER actuators was ~200 µm, which has an effect on properties as introduced in Paper IV. (Dausch 1997, Ounaies et al. 2001, Paper IV) Similarly, the passive ring of the PRESTO actuators has a hindering effect on polarisation. (Paper IV)

In general, the differences between PRESTO actuators and others can be explained by the differences in pre-stress level, stronger clamping of the thicker passive layers of the RAINBOW and THUNDER actuators and the passive ring area introducing high tensile stresses (Paper IV).

Additionally, displacement of the PRESTO actuators closely followed the remanent polarisation as a function of poling fields. This can be used as an indicator of expected displacement properties, as with bulk actuators (Paper IV).

In Paper V, a AgPd pre-stressed PZT 5H PRESTO actuators was used with an additional steel layer. The actuator, with a thickness of 250 µm, produced 71 µm displacement under +2.0 MV/m electric field and 1 Hz frequency. Thicker actuator (410 µm) produced 45 µm displacement under +1.2 MV/m electric field and 1 Hz frequency. Displacement was measured when the actuators were assembled into amplifier structure. Actuators were clamped from a steel layer into the base of the amplifier using screws and connected to the amplifier mechanics by glue. The mechanical amplifier with thinner actuator produced 1.2 mm and 1.15 mm displacements under 0 N and 7.2 mN loads and 2.0 MV/m electric field, respectively. The corresponding amplification factors were ~17 and ~16. The mechanical amplifier with the thicker actuator produced 1.16 mm and 1.08 mm under 0 N and 7.2 mN loads and ~1.2 MV/m electric field, respectively. The inverting amplifier system with a 410 µm thick actuator obtained amplification factors of ~-25 and ~-24. (Paper V)

4.4 Discussion

Optimisation of PRESTO actuators demands further study. Dimensions and mechanical properties of materials, such as thickness (Figs. 7 and 20) and Young’s modulus, have to be carefully considered when designing. This is particularly important in pre-stressed actuators compared to conventional unimorph actuators because the stress state and passive material are created in the same process. From the mechanical point of view, selection of a thicker passive layer can introduce higher displacements due to a more suitable active and passive material ratio, as explained in Chapter 2.2.

In order to estimate the contribution of the mechanical factors, the theoretical maximum axial displacement (Δhmax,circ) was calculated for different PRESTO actuators using equation (2) and parameters in Table 10 (Li et al. 1999).

56

Table 10. Material and driving parameters of the PRESTO actuators used in calculation (Berlincourt et al. 2005, Morgan Electro Ceramics 2005, Sinocera 2005).

Material Diameter [mm]

Thickness [μm]

Young’s modulus [GPa]

Poisson’s ratio

d31 [pm/V] Electric field [MV/m]

PZT 5A 24.5 250 / 375 69 0.35 -171 ± 0.75 PZT 5H 24.5 250 / 375 72 0.34 -274 ± 0.75 Passive 24.5 25-250 37-300 0.30 - -

The effect of the Young’s modulus of the passive layer and the thickness ratio of passive and active material on the displacement is shown in Fig. 20 for a 250 μm thick PZT 5H actuator. The measured displacements of PRESTO actuators have been added from Paper III for comparison.

Fig. 20. Calculated displacement of the PRESTO actuators as a parameter of Young’s modulus of passive layer (Poisson’s ratio 0.3) and ratio of passive layer and PZT. Measured displacements are from 250 μm thick PZT 5H PRESTO actuators with constraining layer driven by ±0.75 MV/m electric field and 10 Hz frequency (Paper III).

Fig. 20 shows that the thickness ratio of the LTCC pre-stressed actuator is near optimum. For the AgPd pre-stressed actuator, a significant increase of the displacement can be expected by introducing a thicker passive layer. Interestingly, measured displacements are much higher than would be expected from the values of the Young’s modulus of silver and LTCC. In order to achieve theoretical displacements equal to the measured one, the AgPd pre-stressing layer should exhibit a Young’s modulus about 115 GPa, as shown in Fig. 20 for PZT 5H actuators. For 250 μm thick PZT 5A PRESTO actuators, calculation suggests that Young’s modulus of the passive AgPd would be ~140 GPa and even higher for thicker actuators. The lowest Young’s modulus which corresponded with measured displacements of the actuators was 37 GPa obtained with a 250 μm thick PZT 5A actuator F4-3 in Paper III. This value was used later in calculations for Young’s modulus of the AgPd layer. Young’s modulus of freely sintered LTCC tape (DuPont 951) is

57

between 120-152 GPa, depending on firing parameters (CTS Microelectronics 2005, DuPont 2005). Constrained sintering is expected to lower these values, but even if 152 GPa Young’s modulus is used for PZT 5H PRESTO actuators, the difference between measured and calculated values is ~40 % with a Poisson’s ratio of 0.3. This value is higher than the difference between measured and theoretical values obtained by Li et al. for unimorph actuators as shown in Table 3 (Li et al. 1999). The same calculation for the different Young’s modulus of passive layer was made for 375 μm thick actuators showing an even higher difference between measured and calculated values.

The value of Poisson’s ratio used in calculations (Fig. 20) was 0.30, corresponding to values of e.g. stainless steel. However, for DuPont 951 LTCC tape, Poisson’s ratio has been previously determined to be 0.17 (CTS Microelectronics 2005). The change applied into calculations increased displacements by 6-15 % over the whole range of thickness ratio and by 11 % at the particular thickness ratio of actuators pre-stressed by LTCC tape.

It should also be noted that in Fig. 20 the Young’s modulus of the piezo ceramics is kept constant. However, Young’s modulus, i.e. inverse elastic compliance, depends on pre-stress and applied electric field (Wang & Cross 1998, Zhang & Zhao 1999, Li et al. 1999). The calculation suggests that using a half of the Young’s modulus of both piezo materials (Table 10) and 37 GPa for the passive layer generates a 15 % increase in maximum displacement compared to the initial Young’s modulus.

All earlier mentioned factors can produce cumulative errors in calculations so that an accurate analysis of material parameters under driving conditions is required in the future. The effective d31 piezoelectric coefficient was calculated for comparison of the PRESTO actuators and other pre-stressed and unimorph actuators. In these calculations Young’s modulus was fixed at 37 GPa and 120 GPa for AgPd and LTCC, respectively. A Poisson’s ratio of 0.3 was used for AgPd and 0.17 for LTCC. The top electrode was ignored in the calculations. Other initial information is in Table 10. Thickness of AgPd, dome heights, measured and calculated displacements, their difference and effective piezoelectric coefficient d31 are calculated for PZT 5A PRESTO actuators in Table 11.

Table 11. Thickness of AgPd, dome heights, measured displacements of different PZT 5A PRESTO actuators under 0.3 N force and ±0.75 MV/m electric field at 10 Hz frequency and their calculated performance.

PZT thickness, label of actuator

Thickness of AgPd [μm]

Height after poling [μm]

Δhmeas [μm]

Δhmax,circ

[μm] (Δhmeas/Δhmax,circ)-1

[%] d31eff

[pm/V] 375 μm

F3-1 1) 45 215 24 12 101 -344 F3-2 2) 33 300 22 9 138 -407 F3-3 1) 35 270 18 10 88 -322 F3-4 1) 24 260 13 7 96 -336

250 μm F4-1 1) 33 665 42 19 119 -374 F4-2 3) 12 220 32 8 304 -690 F4-3 1), 4) 24 360 15 15 0 -171

1) Electrodes sintered first after which pre-stressing layer was sintered. 2) Electrodes and pre-stressing layer sintered at the same time. 3) No sintered top electrode. 4) Reference actuator for determination of Young’s modulus of AgPd.

58

The results of Table 7 and 11 indicate that the thickness of the pre-stressing layer has an effect on dome height and effective d31, since a purely mechanical contribution would give the initial d31=-171 pm/V as the effective value. The effective d31 coefficient of the 375 μm thick actuators in Table 11 corresponded closely with the initial dome height of the actuators (Table 7). This was an expected result as the dome height has been used as an indicator of the pre-stress in the results obtained by others (Mossi et al. 2003). However, when the dome height after poling was compared with the effective d31 coefficient the relation was not as obvious, but still actuators with the highest dome height generated the highest effective d31 coefficient. For thinner actuators the results are clear if actuator F4-2 is ignored. Apparently a constraining top electrode significantly changes the pre-stress state of the actuator. Consequently, dome heights of actuators with different configurations cannot be compared directly. Interestingly, the processing order seems to have a great impact on actuator properties. Actuator F3-2 produced the largest effective d31 value introduced by simultaneous sintering of the top electrode and pre-stressing layer. Also, one firing cycle is avoided which might otherwise diminish the piezoelectric properties due to lead loss (Torah et al. 2004 b, Wang et al. 2004, Yoon et al. 2005).

Measured and calculated displacements, their difference, the effective piezoelectric coefficient d31 and the displacement of the scaled actuator (piezoceramic ∅25x0.25 mm - see Chapter 2.3) using effective d31 in initial and optimal conditions for different PRESTO configurations are listed in Table 12.

Table 12. Measured displacements of different PRESTO actuators with constraining layer under 0.3 N force and ±0.75 MV/m electric field at 10 Hz frequency and their calculated performance.

Actuator type Δhmeas [μm]

Δhmax,circ

[μm] (Δhmeas/Δhmax,circ)-1

[%] d31eff

[pm/V] Δhmax,circ

1) [μm]

Δhmax,circ2)

[μm] PZT 5A 250 μm

F4-1, AgPd 33 μm 42 19 119 -374 44 95 F6-3, LTCC 96 μm 72 57 26 -216 75 75

PZT 5A 375 μm F3-2, AgPd 33 μm 22 9 138 -407 34 105 F6-4, LTCC 96 μm 45 37 21 -207 70 72

PZT 5H 250 μm F5-1, AgPd 33 μm 60 30 101 -552 63 139 F6-1, LTCC 96 μm 117 91 28 -350 122 122

PZT 5H 375 μm F5-3, AgPd 33 μm 51 14 263 -994 80 250 F6-2, LTCC 96 μm 89 59 51 -413 139 144

1) Scaled actuator size (piezoceramic ∅25x0.25 mm) calculated using d31eff and initial thickness ratio– see Chapter 2.3.

2) Scaled actuator size (piezoceramic ∅25x0.25 mm) calculated using d31eff and optimal thickness ratio.

Dome heights were proportional to the effective d31 coefficient when actuator material and thickness were kept the same but the pre-stressing material was changed (Table 7 and

59

12). As shown in Table 12, actuators pre-stressed by AgPd exhibited greater differences between measured and calculated values than did actuators pre-stressed by LTCC. This is probably due to higher tensile pre-stress, (as indicated by the dome height – see Table 7), as a consequence of the larger CTE mismatch between materials, as mentioned earlier. The performance level of the LTCC pre-stressed actuators is between CERAMBOW and co-extruded PZN-PZT actuators (Tables 3 and 12).

PRESTO actuators pre-stressed by AgPd exhibited performance on the level between Prindrex and RAINBOW actuators depending from the configuration (Tables 3, 11 and 12). The highest difference of 304 % between measured and calculated values was obtained with a 250 µm thick PZT 5A PRESTO actuator with a 12 µm thick AgPd pre-stressing layer and silver paint top electrode (Table 11). The highest calculated effective d31=-994 pm/V was obtained with a 375 µm thick PZT 5H PRESTO actuator with a 33 µm thick AgPd pre-stressing layer and AgPd top electrode (Table 12). As suggested by these results, in the optimal case PRESTO actuators would generate the highest displacement among different pre-stressed actuators scaled in size, shape and driving conditions (Table 3, 11 and 12). In practice, displacements up to 250 µm and 175 µm with 250 µm thick (∅25 mm) PZT 5H and 5A actuators respectively (Table 11 and 12) would be obtainable via optimisation. Although, RAINBOW exhibits displacements of the same magnitude (Table 8), the pre-stress level of the PRESTO can be increased and materials with higher Young’s modulus can be applied to further improve their characteristics. However, it is important to notice that these values can decrease or increase significantly resulting from changes in material parameters of the pre-stressing layer used in the model. Therefore detailed characterisation of the pre-stressing passive layer and pre-stressed piezoelectric material under driving conditions will be required in future. Additionally, more experiments are required with different materials and material thicknesses to optimise actuator structure and verify theoretical calculations.

RAINBOW actuators manufactured here produced lower performances (d31eff=-316 pm/V and -191 pm/V for PZT 5H and PZT 5A, respectively) than PRESTO actuators even though they exhibited high displacements with moderate loads (Table 6, 11 and 12). This was due to the non-optimised manufacturing process, with slight differences compared to RAINBOW actuators presented by others. Manufactured thick film actuators showed d31 values of -34.6 pm/V and remanent polarisation of 1.36 μC/cm2, which are orders of magnitude lower than PRESTO actuators. Additionally, pre-stress due to sintering shrinkage is considered to be a harmful effect for piezoceramic in this case. However, these actuators showed quite good characteristics at resonance and thus can be utilised as various mechanical resonators. Furthermore, paste prepared here can be sintered at low temperature and shows promising characteristics for use in functionally graded PRESTO actuators in further experiments.

Electrical characteristics of the PRESTO actuators were discussed in Chapter 4.3 in detail. Increases of the coercive electric field appeared to be at the cost of the remanent polarisation. The relation between remanent polarisation and displacement seems to be valid for pre-stressed actuators in the same way that piezoelectric d33 and d31 coefficients and remanent polarisation are related for bulk actuators (Wang et al. 1991, Paper IV). However, remanent polarisation and piezoelectric d33 or d31 coefficients of PRESTO and bulk actuators are not directly comparable, based on results of effective d31 coefficients (Table 9, 11 and 12). More research is required for the determination of the relation

60

between remanent polarisation and effective d31 coefficient at different stress levels. Poling and structure optimisation of the PRESTO actuators can further improve their displacement capabilities by introducing higher values of effective d31 and coercive electric field (Table 9). (Paper IV)

The 250 µm thick AgPd pre-stressed PZT 5H PRESTO actuator with a glued steel layer, presented in Chapter 4.3, showed a smaller displacement (71 µm) than was expected. (Paper V) Initially, without steel layer, the actuator produced a displacement of ~60 µm under 0.3 N force and ±0.75 MV/m electric field as shown in Table 12. According to equation (2), the displacement is linearly proportional to electric field. Therefore, the actuator should exhibit displacements over 80 µm under +2.0 MV/m electric field and due to the thicker layer of passive material i.e. glue and steel in addition to AgPd. There are several possible reasons for reduced behaviour. Firstly, the actuator was firmly clamped by the steel layer which strongly resisted movement of the piezo actuator. Secondly, the loading effect of the amplifier is not known which might reduce the displacement of a thin actuator. Thirdly, the linear relation of electric field and displacement is not accurate because the displacement of the actuators tends to saturate under strong unipolar electric fields. Finally, the gluing may be ineffective in terms of increasing the displacement and instead may decrease the displacement by resisting the actuation.

The same glued steel layer configuration with a 410 µm thick PZT 5H PRESTO actuator exhibited 45 µm displacement under +1.2 MV/m. Using the same calculation as earlier this actuator should produce 56 µm under ±0.75 MV/m electric field. The closest counterpart is the 375 µm PRESTO actuator with a 33 µm thick AgPd layer (Table 12) which produced a displacement of 51 µm. As can be seen from equation (2) and results shown earlier (Figs. 18 and 19), the displacement of a thicker actuator should be lower than that of a thinner actuator. Therefore it is clear that displacement capabilities are enhanced due to the thicker passive material (additional steel layer), as can be expected based on the unimorph model. The best performance will be obtained when the pre-stress level is set by a AgPd layer and when a steel layer is used to adjust the optimal ratio of passive and active materials. In the model, use of approximately 125 µm and 75 µm thick steel layers would generate the highest displacement for 410 µm and 250 µm thick actuators, respectively. In practice, the highest displacements were obtained by using 100 µm and 50 µm thick steel foils as a consequence of the glue and pre-stressing layers.

As explained in Chapter 2.3, gluing has its problems. However, durability of handling and clamping with other structures is improved by using the solid metal layer beneath the ceramic actuator. In the presented case, the actuator experienced an increased effective d31 coefficient as a consequence of the pre-stress, and gluing of the actuator was carried out after the pre-stressing process. By this means actuators can achieve better durability, better assembling and a higher effective d31 coefficient. In addition, decreased stresses can be expected to be applied to the glue layer. The interlayer of the passive and active material will be subjected to the highest stresses during actuation. Thus using material with a higher Young’s modulus and bonding strength, such as AgPd, instead of glue is beneficial in terms of displacement and endurance. In sections where lower stresses are introduced, adhesives are suitable for improving durability as well as displacement, by mechanical factors as explained earlier. Such a material composite with a graded Young’s modulus requires further studies to be optimised in terms of durability and displacement.

61

Application-specific customisation is often required in spite of relatively large displacements, moderate load bearing capabilities and mass-producible actuators. Customisation can be achieved through flexible manufacturing methods but sometimes it is more feasible and cost effective to make the required changes by external mechanisms, especially if the amount and size of actuators is limited. By using a pre-stressed actuator a high displacement is produced due to the very compact internal leverage. Therefore, the amplification factor and stresses in the external amplifier are reduced and construction and manufacturing are simplified. As a consequence, materials with low mechanical stiffness such as plastic can be used for mechanisms enabling, for example, injection moulding of devices. The amplification mechanism for PRESTO actuators was developed to obtain displacements over 1 mm. This displacement range is very feasible for fine mechanical devices such as low power electromechanical locks, valves, dosing devices and mechanical regulators. In future, amplifier construction will be more compact and efficient with improved load bearing capability utilising optimised pre-stressed actuators.

Further research is required to verify the model for PRESTO actuators. The manufacturing method also enables other options for actuator fabrication, including pre-stressing various sizes and shapes of the actuator as well as pre-stressing patterns. It has already been demonstrated that pre-stressing segments can be realised on the surface of the piezoelectric ceramic so that pre-stressed actuator or sensor arrays can be manufactured on single wafer (Fig. 21 a). It should be also pointed out that presented PRESTO actuators are not functionally graded structures like the PZT/PZT actuator presented by Li et al. (Li et al. 2001). However, such structures are easily manufactured using e.g. piezoelectric paste or tape as a pre-stressing material. Problems associated with limited thickness range of the bulk actuators can be solved with the utilisation, for example, of tape cast piezoelectric structures (Fig. 21 b).

Fig. 21. a) three pre-stressed regions on a piezoelectric disc (∅ 25 mm) b) tape cast piezoelectric actuator (thickness ~70 µm, ∅ 10 mm) with a AgPd pre-stressing layer.

Furthermore, LTCC technology enables a convenient way to build a variety of solid state 3D structures which are durable, hard and inert (Gongora-Rubio et al. 2001, Gongora-Rubio et al. 2004). In order to realise actuators for elevated temperatures a strong and temperature resilient bond between the piezoelectric and another material is required. In

62

addition, temperature durable wiring and electrodes, a high Curie point and enhanced piezoelectric properties are needed. (Schulz et al. 2003) PRESTO actuators offer these possibilities, especially with LTCC. Additionally, they enable realisation of LTCC embedded microwave-driven piezoelectric actuators with high piezoelectric coefficients (Choi et al. 2002, Song et al. 2003). Such actuators and devices with remote control and wireless energy transfer have huge potential markets in various industrial fields and consumer electronics.

5 Conclusions

The main target of this thesis has been to develop novel pre-stressed piezoelectric actuators.

Different PRESTO actuator configurations were realised for 250 µm and 375 μm thick PZT 5A and 5H piezoelectric discs and their displacement characteristics as a function of load were studied. The overall behaviour of the actuators was analogous to RAINBOW and THUNDER actuators. PZT 5A material exhibited better load bearing capability and a 250 µm thick actuator with an LTCC pre-stressing layer increased displacement by 7 % when loaded from 0.3 N to 3 N. A maximum displacement of 118 µm was obtained with a 250 µm thick pre-stressed PZT 5H actuator with a diameter of 25 mm where LTCC tape was utilized as a pre-stressing layer. Similarly, the PZT 5H actuator with same size provided a maximum displacement of 63 µm with a screen-printed pre-stressing AgPd layer. These values show lower displacement capabilities than corresponding RAINBOW and THUNDER actuators. However, optimisation of the pre-stress level and material thicknesses can increase displacement properties by a factor of 2-3. Additionally, a competitive solution is obtained due to the straightforward manufacturing method suitable for mass-production, the integration possibility for LTCC and possibilities for high temperature operation.

The produced mechanical bias and passive layer significantly increased the coercive electric field of the PRESTO actuators. Coercive electric fields up to -1.67 MV/m and -3.02 MV/m were obtained for PZT 5A bulk and PRESTO actuators, respectively. This expands the operational electric field range of the PRESTO actuators, and hence their displacements can be increased. On the other hand, the introduced pre-stress decreased the remanent polarisation of PRESTO actuators compared to that of bulk actuators. This value has been used as an indicator of piezoelectric coefficients d33 and d31. However, since pre-stress also has an influence on piezoelectric coefficients, values of two actuators are not directly related and more research is required in this topic. Mechanical bias enhances the d31 piezoelectric coefficient which is the main contributor for displacement of the bender type PRESTO actuator. The influence of the pre-stress can be seen in the increased displacements. According to the unimorph model, PRESTO actuators pre-stressed by LTCC tape showed moderate enhancement comparable to CERAMBOW actuators i.e. higher than unimorph and some of the most recently

64

developed actuators. On the other hand, PRESTO actuators pre-stressed by AgPd already show the highest effective piezoelectric coefficient d31 (calculated maximum d31eff = -690 pm/V and d31eff = -994 pm/V for PZT 5A and 5H, respectively) among the pre-stressed internally leveraged actuators produced by others. Calculations suggest that displacements up to 250 µm and 175 µm with PZT 5H and 5A actuators (∅25 mm, thickness 250 µm) respectively would be obtainable via optimisation. These theoretical results need to be verified by further experimental results.

RAINBOW and thick film actuators produced here showed significantly lower effective d31 values compared to PRESTO actuators. However, thick film actuators produced a relatively large displacement at resonance with low voltages which can be utilised in low force actuators, resonant devices and sensors with mass-producible manufacturing methods. In further research the paste can also be utilised in a functionally graded PRESTO actuator due to low temperature sintering.

PRESTO actuators were utilised in an external mechanical amplifier producing displacements up to 1.2 mm. In this case the actuators were glued onto steel layers, enhancing the displacement capability due to a thicker layer of passive material. By this means a high effective piezoelectric coefficient d31 could be produced by first pre-stressing and then adjusting the optimal thickness of the passive material. From a practical point of view, such an approach makes structures easy to handle, assemble and attach. Further research on external mechanical amplifiers and optimised pre-stressed structures will improve their displacement, load bearing capability and compactness.

In summary, PRESTO actuators show promising characteristics in terms of large displacement, manufacturability and integration, obtaining the following advantageous features:

− A simple and straightforward manufacturing method with existing mass-production manufacturing technology.

− Utilization of the high piezoelectric coefficients of commercially available materials. − A strong bond due to sintering and a monolithic structure without a separate adhesive

layer. − Increased coercive electric field and effective d31 coefficient. − Integration and miniaturisation possibilities using e.g. LTCC. − A fabrication method enabling segments in the actuator sheet, producing e.g. actuator

matrices. − A actuator structure suitable also for a high temperature operations. − Materials with high Young’s modulus can be utilised as a passive layer increasing

electromechanical properties of the actuator. − Pre-stress level can be easily tuned by selection of materials and thicknesses providing

optimal conditions for different piezoelectric materials. − Functionally graded materials can be utilised.

Properties of the PRESTO actuators can be improved further through optimisation. Remarkable advantages are obtained due to their monolithic, integrated structures and flexible manufacturing methods suitable for mass-production. Embedded solid state actuators can be manufactured, for example, inside the LTCC circuit board in the same process together with a mechanical base, package and electronics. Due to the high chemical resistance of the ceramic and the temperature resilient bonding mechanism,

65

PRESTO actuators offer possibilities for utilisation in harsh environments and elevated temperatures. Wirelessly controlled and microwave driven pre-stressed piezoelectric bender actuators have already been demonstrated in a space application. The concept of integrated actuators, sensors and their matrices with enhanced electromechanical properties and wireless energy transfer is very beneficial and can be utilised in consumer electronics and various industrial fields related, for example, to micro and fine mechanics, telecommunication, instrumentation, fluidics, optics and the car and aerospace industries. Devices which require a high resolution, small size, low power consumption and relatively large displacement are especially attractive, such as dosing devices, electromechanical locks, regulators, positioners, vibrators, speakers, adjusters, pumps, valves, relays, dispensers, micromanipulators and motors, etc.

References

Aoyagi M, Beeby SP & White NM (2002) A novel multi-degree-of-freedom thick-film ultrasonic motor. IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control 49 (2): 151-158.

Barron BW, Li G & Haertling GH (1996) Temperature dependent characteristics of Cerambow actuators. Proc. IEEE International Symposium on Applications of Ferroelectrics, ISAF’96, IEEE, New York, NY, USA, 1: 305–308.

Berlincourt D, Krueger HHA & Near C (2005)Properties of Morgan Electro Ceramic ceramics. Technical publication, Morgan Electro Ceramics Ltd. Cited March 10th 2005 from http://www.morganelectroceramics.com/pdfs/tp226.pdf.

Chandran S, Kugel VD & Cross LE (1997) Crescent: A Novel Piezoelectric Bending Actuator. Proc. International Society for Optical Engineering, SPIE, USA, 3041: 461-469.

Chen HD, Udayakumar KR, Cross LE, Bernstein JJ & Niles LC (1995) Dielectric, ferroelectric, and piezoelectric properties of lead zirconate titanate thick films on silicon substrates. Journal of Applied Physics 77: 3349-3353.

Choi SH, Chu S-H, Song KD & King GC (2002) Microwave-driven multifunctional capability of membrane structures. Journal of American Institute of Aeronautics and Astronautics 2002-5751: 1-7.

Chopra I (2002) Review of state of art of smart structures and integrated systems. Journal of American Institute of Aeronautics and Astronautics 40 (11): 2145-2187.

Chung I, Yoo IK, Lee W, Chung CW, Lee JK & Desu SB (1994) Electrode stress effects on electrical properties of PZT thin film capacitors. Proc. International Symposium on Applications of Ferroelectrics, ISAF’94, IEEE, New York, NY, USA, 527-530.

Colla EL, Thiele ES, Damjanovic D & Setter N (2000) New high performance – low cost monolithic bimorph piezoelectric actuators for applications requiring large displacements with significant forces. Proc. International Symposium on Applications of Ferroelectric, ISAF’2000, IEEE, Piscataway, NJ, USA, 1: 317-320.

Corker DL, Whatmore RW, Ringgaard E & Wolny WW (2000) Liquid-phase sintering of PZT ceramics. Journal of the European Ceramic Society 20: 2039-2045.

CTS Microelectronics (2005) Low-temperature cofired ceramic design and layout guidelines for the fabrication of networks, packages and multichip modules. Cited July 1st 2005 from http://www.ctscorp.com/components/Datasheets/design.pdf.

Culshaw B (1996) Smart Structures and Materials. Artech House, Boston.

67

Dausch DE (1996) The effects of excess PbO addition and composition variation on the processing and properties of tape cast PLZT RAINBOW actuators. Proc. International Symposium on Applications of Ferroelectrics, ISAF’96, IEEE, New York, NY, USA, 2: 771-774.

Dausch DE (1997) Ferroelectric polarization fatique in PZT-based RAINBOWs and bulk ceramics. Journal of American Ceramic Society 80 (9): 2355-2360.

Dogan A, Yoshikawa S & Dosch J (1996) Reliability studies on the bonding materials for ceramic-metal composite actuators. Proc. Symposium of Applications of Ferroelectrics, ISAF’96, IEEE, New York, NY, USA, 2: 747-750.

DuPont Microcircuit materials (2005) 951 Green tape™ datasheet. Cited June 5th 2005 from http://www.mcm.dupont.com/MCM/en_US/index.html.

Elissalde C, Cross LE & Randall CA (1996) Structural-property relations in a reduced and internally biased oxide wafer (RAINBOW) actuator material. Journal of American Ceramic Society 79 (8): 2041-2048.

Furman E, Li G & Haertling GH (1994) Electromechanical properties of Rainbow devices. Proc. International Symposium on Applications of Ferroelectrics, ISAF’94, IEEE, New York, NY, USA, 146-149.

Galassi C, Camporesi G, Fabbri G, Costa AL & Roncari E (2001) Processing of a multilayer bender type actuators. Journal of European Ceramic Society 21: 2011-2014.

Germano C (1971) Flexure mode piezoelectric transducers. IEEE Transactions on Audio and Electroacoustics 19 (1): 6-12.

Gongora-Rubio MR, Espinoza-Vallejos P, Sola-Laguna L & Santiago-Avilés JJ (2001) Overview of low temperature co-fired ceramics tape technology for meso-system technology (MsSt). Sensors and Actuators A 89: 222-241.

Gongora-Rubio MR, Fontes MBA, Mendes da Rocha Z, Richter EM & Angnes L (2004) LTCC manifold for heavy metal detection system in biomedical and environmental fluids. Sensors and Actuators B 103 (1-2): 468-473.

Goodfellow SARL (2005) material properties. Cited June 6th 2005 from www.goodfellow.com. Haertling GH (1994) Rainbow ceramics – a new type of ultra-high-displacement actuator.

American Ceramic Society Bulletin 73 (1): 93-96. Haertling GH (1994) Chemically reduced PLZT ceramics for ultra-high displacement actuators.

Ferroelectrics 154: 101-106. Haertling GH (1997) Rainbow actuators and sensors: a new smart technology. Proc. International

Society for Optical Engineering, SPIE, USA, 3040: 81-92. Hall A, Allahverdi M, Akdogan EK & Safari A (2005) Piezoelectric/electrostrictive multimaterial

PMN-PT monomorph actuators. Journal of the European Ceramic Society 25 (12): 2991-2997. Haris A, Goo NS, Park HC & Yoon KJ (2004) Modeling and analysis for the development of

lightweight piezoceramic composite actuator (LIPCA). Computational Materials Science 30 (3-4): 474-481.

Heyen J, von Kerssenbrock T, Chernyakov A, Heide P & Jacob AF (2003) Novel LTCC/BGA modules for highly integrated millimeter-wave transceivers. IEEE Transactions on Microwave Theory and Techniques 51 (12): 2589-2596.

Holc J, Hrovat M & Kosec M (1999) Interactions between alumina and PLZT thick films. Materials Research Bulletin 34 (14-15): 2271-2278.

Hooker MW (1998) Properties of PZT-based piezoelectric ceramic between -150 and 250 °C. Technical report, NASA, Hanover, MD, USA, NASA/CR-1998-208708: 1-30.

Hrovat M, Holc J, Drnovšek S, Belavič D, Cilenšek J & Kosec M (2006) PZT thick films on LTCC substrates with an interposed alumina barrier layer. Journal of the European Ceramic Society 26 (6): 897-900.

68

Hughes WJ & Bednarchik PG (2000) Electrical biasing of PZT ceramic material. Proc. International Symposium on Applications of Ferroelectrics, ISAF-2000, IEEE, Piscataway, NJ, USA, 1: 41-44.

Iborra E, Olivares J, Clement M, Vergara L, Sanz-Hervás A & Sangrador J (2004) Piezoelectric properties and residual stress of sputtered AIN thin films for MEMS applications. Sensors and Actuators A 115: 501-507.

Jantunen H, Kangasvieri T, Vähäkangas J & Leppävuori S (2003) Design aspects of microwave components with LTCC technique. Journal of European Ceramic Society 23 (14): 2541-2548.

Juuti J, Moilanen H & Leppävuori S (2001) Manufacturing of rainbow actuators for displacement applications. Proc. euspen International Conference, euspen, Cranfield Univ., Bedford, UK, 1: 204-207.

Juuti J, Heinonen E & Leppävuori S (2004) RAINBOW and tape casting ceramic technologies in realisation of miniature actuators and motors. Proc. world congress in Mechanism and Machine Science (IFToMM), China Machinery Press, Tianjin, China, 3: 1142-1445.

Kim KY, Park KH, Park HC, Goo NS & Yoon KJ (2005) Performance evaluation of lightweight piezo-composite actuators. Sensors and Actuators A 120: 123-129.

Koch M, Harris N, Evans AGR, White NM & Brunnschweiler A (1998) A novel micromachined pump based on thick-film piezoelectric actuation. Sensors and Actuators A 70: 98-103.

Kugel VD, Chandran S & Cross LE (1997) Comparative analysis of piezoelectric bending-mode actuators. Proc. International Society for Optical Engineering, SPIE, USA, 3040: 70-80.

Lappalainen J, Frantti J, Moilanen H & Leppävuori S (1995) Excimer laser ablation of PZT thin films on silicon cantilever beams. Sensors and Actuators A 46-47: 104-109.

Lappalainen J (1999) Laser-ablation deposition and characterization of polycrystalline Nd-modified Pb(Zr,Ti)O3 thin films. Acta Universitatis Ouluensis C 163.

Lee S, Jang S-H, Kim YS, Park HC & Yoon KJ (2005) Material nonlinear characteristics of the 3203HD PZT wafer under high electric fields. Sensors and Actuators A 121: 450-456.

Leppävuori S, Moilanen H, Uusimäki A & Kruusing A (1999) Preparation and applications of printed thick PZT films on different substrates. Proc. European Microelectronics and Packaging Conference, IMAPS Europe, Cambridge, UK, 48-54.

Li G, Furman E & Haertling GH (1997) Stress-enhanced displacements in PLZT Rainbow actuators. Journal of American Ceramic Society 80 (6): 1382-1388.

Li J-F, Takagi K, Ono M, Pan W, Watanabe R, Almajid A & Taya M (2003) Fabrication and evaluation of porous piezoelectric ceramics and porosity-graded piezoelectric actuators. Journal of American Ceramic Society 86 (7): 1094-1098.

Li X, Shih WY & Aksay IA (1999) Electromechanical behaviour of PZT-brass unimorphs. Journal of American Ceramic Society 82 (7): 1733-1740.

Li X, Vartuli JS, Milius DL, Aksay IA, Shih WY & Shih W-H (2001) Electromechanical properties of a ceramic d31-gradient flextensional actuator. Journal of American Ceramic Society 84 (5): 996-1003.

Li X, Shih WY, Vartuli JS, Milius DL, Aksay IA & Shih W-H (2002) Effect of a transverse tensile stress on the electric-field-induced domain reorientation in soft PZT: in situ XRD study. Journal of American Ceramic Society 85 (4): 844-850.

Lynch CS (1996) The effect of uniaxial stress on the electro-mechanical response of 8/65/35 PLZT. Acta Materialia 44 (10): 4137-4148.

Moilanen H & Leppävuori S (1993) Determination of low-voltage actuator displacements in d.c. and a.c. electric fields using an optical interferometer. Sensors and Actuators A 37-38: 555-560.

Moilanen H & Leppävuori S (2001) Laser interferometric measurements of displacement-field characteristics of piezoelectric actuators and actuator materials. Sensors and Actuators A 92: 236-334.

69

Morgan Electro Ceramics Ltd. (2005) Useful relationships for circular ceramic bender bimorphs. Technical publication, Morgan Electro Ceramics Ltd. Cited June 5th 2005 from http://www.morganelectroceramics.com/pdfs/tp230.pdf.

Mossi K, Ounaies Z, Smith R & Ball B (2003) Pre-stressed curved actuators: characterization and modeling of their piezoelectric behaviour. Technical report of CRSC, North Carolina State University, Raleigh, NC, USA, CRSC-TR03-17: 1-13. Cited April 2nd 2005 from http://www.ncsu.edu/crsc/reports/ftp/pdf/crsc-tr03-17.pdf.

Mossi KM, Selby GV & Bryant RG (1998) Thin-layer composite unimorph ferroelectric driver and sensor properties. Materials Letter 35 (1-2): 39-49.

Moulson AJ & Herbert JM (2003) Electroceramics. John Wiley & Sons Ltd., Chichester, England. Nakamura K, Ando H & Shimizu H (1987) Ferroelectric domain inversion caused in LiNbO3

plates by heat treatment. Applied Physics Letters 50 (20): 1413-1414. Niezrecki C, Brei D, Balakrishnan S & Moskalik A (2001) Piezoelectric actuation: state of the art.

The Shock and Vibration Digest 33 (4): 269-280. Nolan-Proxmire D & Henry K (1996) NASA rolls out award-winning “THUNDER”. NASA news

release, NASA, Washington, DC, USA, no. 96-154. Cited July 28th 2005 from http://www.nasa.gov/centers/langley/news/releases/1996/Oct96/96_154.html.

Ounaies Z, Mossi K, Smith R & Bernd J (2001) Low-field and high-field characterization of THUNDER actuators. Proc. International Society for Optical Engineering, SPIE, USA, 4333: 399-407.

Paradise WE, Wang F & Haertling GH (1994) Effects of applied stress on the dielectric properties of PLZT thin films. Proc. International Symposium on Applications of Ferroelectrics, ISAF’94, IEEE, New York, NY, USA, 784-787.

Park SE & Shrout TR (1997) Ultrahigh strain and piezoelectric behaviour in relaxor based ferroelectric single crystals. Journal of Applied Physics 82: 1804-1811.

Pearce TW & Button DH (1998) Alternative processing routes for RAINBOW-type piezoelectric devices. Proc. International Symposium on Applications of Ferroelectrics, ISAF’98, IEEE, Piscataway, NJ, USA, 547-549.

PI Ceramic GmbH (2005) Actuator catalogue. Cited June 28th 2005 from http://www.piceramic.de/download.html.

Pokharel BP & Pandey D (1999) Irreversibility of the antiferroelectric to ferroelectric phase transition in (Pb0.90Ba0.10)ZrO3 ceramics. Journal of Applied Physics 86: 3327-3332.

Sawyer CB (1931) Piezoelectric device. US Patent 1,802,782. Schulz MJ, Sundaresan MJ, McMichael J, Clayton D, Sadler R & Nagel B (2003) Piezoelectric

materials at elevated temperature. Journal of Intelligent Material Systems and Structures 14: 693-705.

Schwartz RW, Cross LE & Wang QM (2001) Estimation of the effective d31 coefficients of the piezoelectric layer in rainbow actuators. Journal of American Ceramic Society 84 (11): 2563-2569.

Schwartz RW & Moon Y-W (2001) Domain configuration and switching contributions to the enhanced performance of Rainbow actuators. Proc. International Society for Optical Engineering, SPIE, USA, 4333: 408-417.

Shepard JF Jr, Trolier-McKinstry S, Hendrickson MA & Zeto R (1996) Properties of PZT thin films as a function of in-plane biaxial stress. Proc. International Symposium on Applications of Ferroelectrics, ISAF’96, IEEE, New York, NY, USA, 1: 161-165.

Sherrit S, Wiederick HD, Mukherjee BK & Haertling GH (1994) Dielectric, piezoelectric and hydrostatic properties of PLZT based RAINBOW ceramics. Proc. International Symposium on Applications of Ferroelectrics, ISAF’94, IEEE, New York, NY, USA, 390-393.

Shih WY, Shih W-H & Aksay IA (1997) Scaling analysis for the axial displacement and pressure of flextensional transducers. Journal of American Ceramic Society 80 (5): 1073-1078.

70

Simon L, Le Dren S & Gonnard P (2001) PZT and PT screen-printed thick films. Journal of the European Ceramic Society 21: 1441-1444.

Sinocera (2005) material datasheet. Cited August 1st 2005 from http://www.sinocera.net. Song KD, Yi WJ, Chu S-H & Choi SH (2003) Microwave-driven THUNDER materials.

Microwave and Optical Technology Letters 36 (5): 331-333. Takagi K, Li J-F, Yokoyama S & Watanabe R (2003) Fabrication and evaluation of PZT/Pt

piezoelectric composites and functionally graded actuators. Journal of the European Ceramic Society 23: 1577-1583.

Taya M, Almajid AA, Dunn M & Takahashi H (2003) Design of bimorph piezo-composite actuators with functionally graded microstructure. Sensors and Actuators A 107: 248-260.

Tenzer PE & Mrad RB (2004) On amplification in inchworm™ precision positioners. Mechatronics 14: 515-531.

Torah RN, Beeby SP & White NM (2004) Experimental investigation into the effect of substrate clamping on the piezoelectric behaviour of thick-film PZT elements. Journal of Physics D 37: 1-5.

Torah RN, Beeby SP & White NM (2004) Improving the piezoelectric properties of thick-film PZT: the influence of paste composition, powder milling process and electrode material. Sensors and Actuators A 110: 378-384.

Uchino K, Yoshizaki M, Kasai K, Yamamura H, Sakai N & Asakura H (1987) “Monomorph Actuators” using semiconductive ferroelectrics. Japanese Journal of Applied Physics 26 (7): 1046-1049.

Uchino K (1997) Piezoelectric actuators and ultrasonic motors. Kluwer Academic Publisher, Boston.

Yang G, Liu S-F, Ren W & Mukherjee BK (2000) Uniaxial stress dependence of the piezoelectric properties of lead zirconate titanate ceramics. Proc. International Symposium on Applications of Ferroelectrics, ISAF 2000, IEEE, Piscataway, NJ, USA, 1: 431-434.

Yang G, Ren W, Liu S-F, Masys AJ & Mukherjee BK (2000) Effects of uniaxial stress and DC bias field on the piezoelectric, dielectric, and elastic properties of piezoelectric ceramics. Proc. IEEE Ultrasonics Symposium, IEEE, Piscataway, NJ, USA, 2: 1005-1008.

Yoon KJ, Park KH, Park HC & Perreux D (2003) Thermal deformation analysis of curved actuator LIPCA with a piezoelectric layer and fiber composite layers. Composites Science and Technology 63 (3-4): 501-506.

Yoon C-B, Lee S-M, Lee S-H & Kim H-E (2005) PZN-PZT flextensional actuator by co-extrusion process. Sensors and Actuators A 119: 221-227.

Yoon C-B, Lee S-H, Lee S-M & Kim H-E (2006) Multilayer bender-type PZT-PZN actuator by co-extrusion process. Journal of the European Ceramic Society in press.

Walter V, Delobelle P, Le Moal P, Joseph E & Collet M (2002) A piezo-mechanical characterization of PZT thick films screen-printed on alumina substrate. Sensors and Actuators A 96: 157-166.

Wang CH, Chang SJ & Chang PC (2004) Effect of sintering conditions on characteristics of PbTiO3-PbZrO3-Pb(Mg1/3Nb2/3)O3-Pb(Zn1/3Nb2/3)O3. Materials Science and Engineering B 111: 124-130.

Wang H, Newnham RE, Cross LE & Pan WY (1991) Piezoelectric properties of PLZT ceramics at different remanent polarization levels. IEEE Seventh International Symposium on Applications of Ferroelectrics, IEEE, New York, NY, USA, 422-425.

Wang H-W, Cheng S-Y & Wang C-M (1989) Optimization of poling process for piezoelectric PZT ceramics. Proc. Japan International Electronic Manufacturing Technology Symposium, Japan IEMT, IEEE, New York, NY, USA, 263-266.

Wang Q-M & Cross LE (1998) Determination of Young’s modulus of the reduced layer of a piezoelectric RAINBOW actuator. Journal of Applied Physics 83 (10): 5358-5363.

71

Wang Q-M & Cross LE (1999) Analysis of high temperature reduction processing of RAINBOW actuator. Material Chemistry and Physics 58: 20-25.

Wang Q-M & Cross LE (1999) Tip deflection and blocking force of soft PZT-based cantilever RAINBOW actuators. Journal of American Ceramic Society 82 (1): 103-110.

Wang Q-M, Zhang Q, Xu B, Liu R & Cross LE (1999) Nonlinear piezoelectric behaviour of ceramic bending mode actuators under strong electric fields. Journal of Applied Physics 86 (6): 3352-3360.

Wise SA (1998) Displacement properties of RAINBOW and THUNDER piezoelectric actuators. Sensors and Actuators A 69: 33-38.

Wolny WW (2000) Piezoceramic thick films – technology and applications state of the art in Europe. Proc. International Symposium on Applications of Ferroelectrics, ISAF 2000, IEEE, Piscataway, NJ, USA, 1: 257-262.

Zhang QM & Zhao J (1999) Electromechanical properties of lead zirconate titanate piezoceramics under the influence of mechanical stresses. IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control 46 (6): 1518-1526.

Zhao J & Zhang QM (1996) Effect of mechanical stress on the electromechanical performance of PZT and PMN-PT ceramics. Proc. International Symposium on Applications of Ferroelectrics, ISAF’96, IEEE, New York, NY, USA, 2: 971-974.

Zhao J, Zhang QM & Mueller V (1998) Uniaxial stress dependence of electromechanical and dielectric properties of 0.9PMN-0.1PT electroceramics. Proc. International Symposium on Applications of Ferroelectrics, ISAF’98, IEEE, Piscataway, NJ, USA, 361-364.

Zhou D, Kamlah M & Munz D (2005) Effects of uniaxial pre-stress on the ferroelectric hysteretic response of soft PZT. Journal of the European Ceramic Society 25: 425-432.

Zhu X & Meng Z (1995) Operational principle, fabrication and displacement characteristics of a functionally gradient piezoelectric ceramic actuator. Sensors and Actuators A 48: 169-176.

Original publications

I Juuti J, Heinonen E, Moilanen V-P & Leppävuori S (2004) Displacement, stiffness and load behaviour of laser-cut RAINBOW actuators. Journal of the European Ceramic Society 24: 1901-1904.

II Juuti J, Lozinski A & Leppävuori S (2004) LTCC compatible PLZT thick-films for piezoelectric devices. Sensors and Actuators A 110: 361-364.

III Juuti J, Jantunen H, Moilanen V-P & Leppävuori S (2006) Manufacturing of pre-stressed piezoelectric actuators by post-fired biasing layer. Accepted for publication to IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control in December, 2005.

IV Juuti J, Jantunen H, Moilanen V-P & Leppävuori S (2005) Poling conditions of pre-stressed piezoelectric actuators and their displacement. Journal of Electroceramics 15: 57-64.

V Juuti J, Kordás K, Lonnakko R, Moilanen V-P & Leppävuori S (2005) Mechanically amplified large displacement piezoelectric actuators. Sensors and Actuators A 120: 225-231.

Published papers have been reprinted with premission by the respective copyright holder.

I © 2006, Elsevier B.V. II © 2006, Elsevier B.V. III © 2006, IEEE IV © 2006, Springer Science + Business Media, Inc. V © 2006, Elsevier B.V.

Original publications are not included in the electronic version of the dissertation.

A C T A U N I V E R S I T A T I S O U L U E N S I S

Distributed byOULU UNIVERSITY LIBRARY

P.O. Box 7500, FI-90014University of Oulu, Finland

Book orders:OULU UNIVERSITY PRESSP.O. Box 8200, FI-90014University of Oulu, Finland

S E R I E S C T E C H N I C A

218. Sfakiotaki, Despina (2005) Analysis of movement in sequential space. Perceivingthe traditional Japanese tea and stroll garden

219. Phan, Vinh V. (2005) Smart packet access and call admission control for efficientresource management in advanced wireless networks

220. Ylianttila, Mika (2005) Vertical handoff and mobility — system architecture andtransition analysis

221. Typpö, Asser (2005) Pellon alarajan muutos ja sen vaikutukset viljelyyn jaympäristöön Keski-Pohjanmaalla ja Pohjois-Pohjanmaan eteläosassa

222. Jokelainen, Janne (2005) Hirsirakenteiden merkitys asema-arkkitehtuurille 1860–1950

223. Hadid, Abdenour (2005) Learning and recognizing faces: from still images to videosequences

224. Simola, Antti (2005) Turvallisuuden johtaminen esimiestyönä. Tapaustutkimuspitkäkestoisen kehittämishankkeen läpiviennistä teräksen jatkojalostustehtaassa

225. Pap, Andrea Edit (2005) Investigation of pristine and oxidized porous silicon

226. Huhtinen, Jouni (2005) Utilization of neural network and agent technologycombination for distributed intelligent applications and services

227. Ojala, Satu (2005) Catalytic oxidation of volatile organic compounds andmalodorous organic compounds

228. Sillanpää, Mervi (2005) Studies on washing in kraft pulp bleaching

229. Lehtomäki, Janne (2005) Analysis of energy based signal detection

230. Kansanen, Kimmo (2005) Wireless broadband single-carrier systems with MMSEturbo equalization receivers

231. Tarkkonen, Juhani (2005) Yhteistoiminnan ehdoilla, ymmärryksen ja vallanrajapinnoilla. Työsuojeluvaltuutetut ja -päälliköt toimijoina, työorganisaatiotyhteistoiminnan areenoina ja työsuojelujärjestelmät kehittämisen kohteina

232. Ahola, Timo (2005) Intelligent estimation of web break sensitivity in papermachines

233. Karvonen, Sami (2006) Charge-domain sampling of high-frequency signals withembedded filtering

234. Laitinen, Risto (2006) Improvement of weld HAZ toughness at low heat input bycontrolling the distribution of M-A constituents

C235etukansi.fm Page 2 Friday, March 3, 2006 3:24 PM

ABCDEFG

UNIVERS ITY OF OULU P .O . Box 7500 F I -90014 UNIVERS ITY OF OULU F INLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

S E R I E S E D I T O R S

SCIENTIAE RERUM NATURALIUM

HUMANIORA

TECHNICA

MEDICA

SCIENTIAE RERUM SOCIALIUM

SCRIPTA ACADEMICA

OECONOMICA

EDITOR IN CHIEF

EDITORIAL SECRETARY

Professor Mikko Siponen

Professor Harri Mantila

Professor Juha Kostamovaara

Professor Olli Vuolteenaho

Senior assistant Timo Latomaa

Communications Officer Elna Stjerna

Senior Lecturer Seppo Eriksson

Professor Olli Vuolteenaho

Publication Editor Kirsti Nurkkala

ISBN 951-42-7988-3 (Paperback)ISBN 951-42-7989-1 (PDF)ISSN 0355-3213 (Print)ISSN 1796-2226 (Online)

U N I V E R S I TAT I S O U L U E N S I SACTAC

TECHNICA

OULU 2006

C 235

Jari Juuti

PRE-STRESSED PIEZOELECTRIC ACTUATOR FOR MICRO AND FINE MECHANICAL APPLICATIONS

FACULTY OF TECHNOLOGY,DEPARTMENT OF ELECTRICAL AND INFORMATION ENGINEERING,INFOTECH OULU,UNIVERSITY OF OULU

C 235

AC

TA Jari Juuti

C235etukansi.fm Page 1 Friday, March 3, 2006 3:24 PM